Pith. sign in

REVIEW 4 major objections 5 minor 101 references

RadarSeq: A Temporal Vision Framework for User Churn Prediction via Radar Chart Sequences

T0 review · 4 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that rendering each courier's daily behavior as a radar-chart image and reading the 50-day sequence with a CNN plus bidirectional LSTM predicts courier churn with an F1 of 0.847, beating a static radar-chart ViT baseline…

desk verdict A plausible applied churn model whose headline result only becomes meaningful if the authors fix the temporal boundary between the input window and the churn-defining inactivity period; it deserves a serious referee because that is an addressable reporting gap. read the letter →

arxiv 2506.17325 v1 pith:4N4VRTSY submitted 2025-06-18 cs.CV cs.AI

Sina Najafi , M. Hadi Sepanj , Fahimeh Jafari This is my paper · ORCID
classification cs.CVcs.AI
keywords customerchurnpredictiontemporalmodelingcomputervisionradarchartsLSTMdeeplearningdeliverycourier
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper claims that churn among gig-economy couriers can be predicted much more accurately by preserving the temporal order of daily behavior than by summarizing it into a single static picture. The method turns each courier's daily feature vector into a small radar-chart image and feeds the resulting 50-image sequence into a pretrained CNN followed by a bidirectional LSTM. On a real dataset of about 16,000 couriers, the paper reports F1 of 0.847, precision of 0.884, ROC-AUC of 0.981, and MCC of 0.71, outperforming both classical tabular models and a static radar-chart Vision Transformer baseline by large margins. The central message is that the temporal unfolding of behavior carries information that static summaries throw away, and that radar charts provide a visual encoding on which pretrained image models work well. If correct, this makes churn prediction in non-subscription platforms a spatio-temporal vision problem rather than a tabular classification problem.

What carries the argument

The machinery is a daily radar chart sequence: each courier's day is rendered as a polar radar chart image, constructed from a d-dimensional behavioral feature vector, and the 50-day sequence is processed by a CNN encoder into per-day embeddings. A bidirectional LSTM then compresses the ordered embeddings into a single representation, which a small MLP classifies with a sigmoid. The radar chart supplies a visual inductive bias that lets an ImageNet-pretrained CNN read multivariate behavioral features, while the bidirectional LSTM supplies the temporal inductive bias that the paper identifies as the main source of the performance gain.

What would settle it

Inspect the dataset's date ranges: if any day of the 50-day input window falls inside the 45-day inactivity period that defines the label, then the high ROC-AUC reflects detecting existing inactivity rather than predicting future churn, and re-running with the input window forced to end before the inactivity period would settle which interpretation is correct.

Watch

Extended reading notes

Core claim

The central claim is that preserving temporal structure as a sequence of daily radar chart images, rather than collapsing behavior into one aggregate image, is what drives accurate churn prediction in a setting without explicit cancellation signals. The model encodes each day's feature vector into a 32-by-32 grayscale radar chart, extracts per-day embeddings with a truncated pretrained MobileNetV2, and passes the ordered embeddings through a two-layer bidirectional LSTM followed by a binary classifier. The paper reports that this CNN plus LSTM design reaches an F1 of 0.847 and a ROC-AUC of 0.981, beating a static radar-chart ViT by 17.7 F1 points. Ablations reported in the paper show that removing the LSTM costs 6.9 ROC-AUC points, while replacing radar images with raw tabular features costs about 8 ROC-AUC points, which the authors use to argue that both temporal modeling and the visual encoding contribute to the gain. In the authors' telling, the improvement demonstrates that churn is a process that unfolds over time and is learnable from visual sequences of daily behavior.

Load-bearing premise

The load-bearing premise is that the 50-day input window ends before the 45-day inactivity period that defines churn begins, so the radar charts of churners do not already display the inactivity being predicted; the paper never states the exact gap.

Editorial extensions

If this is right

  • If the central claim is correct, static single-image churn models leave a large performance margin on the table, and temporal visual sequences become the stronger baseline for behavior-based churn prediction in non-subscription settings.
  • The modular design implies the same CNN plus LSTM pipeline can be retrained for other behavioral log domains, such as e-commerce or ride-hailing, by swapping the feature set and radar rendering.
  • Because labels come from a business-defined 45-day inactivity rule rather than clustering-based pseudo-labels, the approach provides cleaner supervision than the ViT baseline it outperforms.
  • The reported engineering choices, including 32-by-32 images, offline resizing, mixed precision, and gradient accumulation, imply the method can be deployed at scale, processing roughly 100,000 sequences per epoch in under two hours on a single GPU.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A direct test of true forecasting would require the input window to end before the 45-day inactivity period that defines the label; because the paper only says windows are 'aligned to precede' the threshold, part of the reported AUC could reflect detecting already-inactive couriers rather than predicting future churn.
  • The same pipeline could be evaluated with a lead-time sweep, predicting churn 7, 14, or 30 days before the inactivity window begins, which would quantify how early the temporal signals are visible; the current binary setup does not report this.
  • The authors' framing predicts that any image encoding preserving daily order should retain much of the CNN plus LSTM advantage, so a comparison against Gramian angular fields or heatmaps would test whether radar charts specifically, or temporal ordering generally, is the load-bearing component.
  • Because several radar features are themselves 7-day averages and week-over-week changes, some temporal signal already lives inside each daily chart; an ablation using only same-day raw features would separate cross-day dynamics from within-feature smoothing.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper proposes RadarSeq, a churn-prediction framework for gig-economy couriers that encodes each user's daily behavioral features as a 32×32 radar-chart image and feeds the resulting 50-image sequence into a pretrained MobileNetV2 encoder followed by a bidirectional LSTM. Churn labels are defined by a 45-day inactivity rule, and the method is evaluated on a private dataset of roughly 16,000 bikers. The authors report F1 0.847, precision 0.884, ROC-AUC 0.981, and MCC 0.71, outperforming CatBoost, ChurnViT, CNN-only, and CNN+MLP baselines, with ablations attributing part of the gain to temporal modeling and to radar-image encoding. The paper also discusses interpretability via SHAP and Grad-CAM, and deployment-oriented optimizations.

Significance. If the reported results hold under a clean temporal separation between the input window and the label-defining inactivity period, the paper makes a useful empirical contribution: it demonstrates a concrete way to apply pretrained vision backbones to a tabular-sequence problem while preserving temporal structure, and it provides ablations showing the contribution of the recurrent component. The business-rule labeling is a sensible alternative to clustering-based pseudo-labels. However, the empirical claims are currently supported only by a single split of an unnamed dataset with no confidence intervals, and the most load-bearing methodological detail—the exact alignment between the 50-day input window and the 45-day churn window—is left unspecified, so the central temporal-dynamics interpretation is not yet established.

major comments (4)
  1. [§3.5, §4.1] The temporal alignment between the input window and the churn label is unspecified. Section 3.5 says the 50-day window is "aligned to precede the 45-day churn threshold," and Section 4.1 says only that labels are assigned by a 45-day inactivity rule, but neither section states the exact number of days between the last input day and the first day of the inactivity period. If any of the 50 input days falls inside the 45-day inactivity span, then churners will have empty or white radar charts for those days, and the CNN+BiLSTM can detect inactivity rather than predict future churn, which would explain the very high ROC-AUC of 0.981 and the gain over static baselines. Please state the exact cutoff, verify that no label information appears inside the input window, and report the fraction of churners with zero-activity days within the window; if the current pipeline does not enforce a strict gap, re-run the evaluation with such a gap.
  2. [§4.2, Table 3] Section 4.2 promises a comparison against "classical LSTM, Transformer, and Random Forest models," but Table 3 contains only CatBoost, ChurnViT, CNN-Only, CNN+MLP, and the proposed model. The text's claim that "our method significantly outperforms all baselines" is therefore not verifiable from the reported table. Please add the promised baseline rows with the same metrics, or remove the claim about those baselines.
  3. [Table 3, §4.3] All quantitative results are point estimates from a single user-level split and a single training run; no confidence intervals, standard deviations across seeds, or significance tests are reported. Since the headline contributions are stated as very large gains (+17.7 F1, +29.4 precision, +16.1 AUC), the absence of variance estimates makes it impossible to judge whether the ranking is stable. Please report results over multiple random seeds (and ideally multiple temporal cutoffs) and state whether the differences are statistically significant.
  4. [§4.3, Table 3] The raw-features ablation is reported only as "reduces AUC to 0.901" in the text, but the corresponding row is missing from Table 3, and the raw-feature setup is not described (which features, how they are normalized, what sequence length, what model size). Without this configuration detail, the claim that radar-image encoding adds value beyond raw tabular features cannot be assessed. Please add the full ablation result and its experimental configuration.
minor comments (5)
  1. [Table 1] In Table 1, the ChurnViT row indicates "No" for both Temporal and Visual, but the same row lists "Radar Img" as the input type; the Visual column should be "Yes" for ChurnViT, since the method converts customer features into radar chart images.
  2. [§3.5] The statement that zero-padded white radar charts are prepended for short histories implicitly assumes that a blank chart semantics equals "no activity." This assumption should be stated explicitly and validated, especially because short histories may themselves correlate with churn.
  3. [§4.1] The dataset is described only as "approximately 16,000 bikers with daily transactional logs" and is not named or released. Please provide at least the collection time period, the platform context, the class imbalance ratio, and a description of the feature extraction from raw logs to enable reproducibility.
  4. [§1] The phrase "the first known attempt" is a strong claim that is difficult to verify; it would be safer to say "to the best of our knowledge" and cite the most recent related work on temporal visual encodings of user behavior.
  5. [§4.3] The discussion says that the confusion matrix in Figure 3 "shows 81.93% true negatives and 13.83% true positives," but the figure caption says "Epoch 20's confusion matrix," which suggests the numbers are from the test set at epoch 20; please clarify whether the confusion matrix is from the validation or test set and whether early stopping at epoch 8 was used for the final model.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: RadarSeq's reported gains are empirical comparisons against external baselines; the central derivation is self-contained, with only a temporal-boundary reporting gap.

full rationale

The paper's central claim is that a CNN+BiLSTM over daily radar chart images outperforms static baselines. The reported metrics (F1 0.847, precision 0.884, ROC-AUC 0.981) are measured on a held-out user-level test split and are not identities, fitted constants, or renamed inputs. The churn label is a business rule (45-day inactivity) that is external to the model's own outputs, and Section 3.5 states the 50-day input window is 'aligned to precede the 45-day churn threshold,' so as written the target is not definitionally contained in the inputs. No parameter is fitted to the test set and then presented as a prediction; the ablations and baseline comparisons are empirical. The only self-citations (refs 72 and 73) concern self-supervised learning and multi-target tracking and are not load-bearing for the churn claim. The main caveat is that the paper does not specify the exact offset between the last day in the input window and the first day of the 45-day inactivity period; if the window overlapped the label period, the radar charts for churners would encode inactivity directly, making the high AUC a recognition task rather than a prediction task. However, the paper's explicit statement that the window precedes the threshold contradicts that overlap, and the hard rule against speculative circularity prevents treating an unspecified but denied overlap as a demonstrated circular step. This is a correctness/reporting risk, not a circularity finding.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

No new physical, formal, or model-level entities are invented. The radar chart sequence is a data representation, and the Activity Score and other engineered features are derived features, not independent postulated entities. The main unstated cost is temporal alignment: the 45-day label and the 50-day input window must be disjoint for the results to mean prediction. Hyperparameters and preprocessing choices are numerous and largely unablated.

free parameters (6)
  • 45-day churn inactivity threshold = 45 days
    Hand-chosen label definition based on operational insight; every label and every reported metric depends on it, with no sensitivity analysis.
  • Input window length T = 50 days
    Chosen sequence length aligned to the 45-day threshold; no ablation on window length is reported.
  • Sliding window strides = 5 (coarse) and 1 (fine)
    Set by hand to control sequence count and temporal diversity; no analysis of the effect on results.
  • Radar image resolution = 32 by 32 grayscale
    Downsampled from 224 by 224 for efficiency; no resolution ablation is provided.
  • LSTM hidden size and learning rate = h=128, lr=1e-4
    Standard hyperparameters chosen without a reported sensitivity study.
  • Checkpoint selection epoch = epoch 8 (highest validation AUC)
    The retained checkpoint is selected by peeking at validation AUC; the text does not describe a fully independent test set protocol beyond the single split.
assumptions (5)
  • domain assumption A courier inactive for 45 consecutive days is a churner.
    Used as ground truth in Section 3.7; not validated against alternative labels, business outcomes, or other thresholds.
  • domain assumption The 50-day input window ends before the 45-day inactivity period begins.
    Section 3.5 says windows are 'aligned to precede' the threshold, but the exact cutoff is never specified. Without this condition, the result is detection of ongoing inactivity rather than prediction of future churn.
  • domain assumption Radar chart images render daily behavioral features in a way that a CNN pretrained on ImageNet can exploit.
    Assumed in Sections 1 and 3.3; the raw-features LSTM ablation is the only supporting evidence and it is not a direct test of the visual encoding assumption.
  • ad hoc to paper White radar charts used for zero padding carry the same semantics as no activity.
    Section 3.5 prepends white charts for shorter histories without showing that this choice does not distort predictions or introduce artifacts.
  • standard math Binary cross-entropy minimization is the correct objective for this deployment.
    Section 3.2 defines the loss, but the asymmetric costs of false positives and false negatives in retention campaigns are not embedded in the loss or in a decision threshold.

how reviews work

0 comments
Cite this review

Pith. "Pith review of RadarSeq: A Temporal Vision Framework for User Churn Prediction via Radar Chart Sequences." pith.science (2026). https://pith.science/paper/4N4VRTSY

@misc{pith2026250617325,
  author       = {Pith},
  title        = {Pith review of: RadarSeq: A Temporal Vision Framework for User Churn Prediction via Radar Chart Sequences},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4N4VRTSY}},
  note         = {Machine review of arXiv:2506.17325}
}
read the original abstract

Predicting user churn in non-subscription gig platforms, where disengagement is implicit, poses unique challenges due to the absence of explicit labels and the dynamic nature of user behavior. Existing methods often rely on aggregated snapshots or static visual representations, which obscure temporal cues critical for early detection. In this work, we propose a temporally-aware computer vision framework that models user behavioral patterns as a sequence of radar chart images, each encoding day-level behavioral features. By integrating a pretrained CNN encoder with a bidirectional LSTM, our architecture captures both spatial and temporal patterns underlying churn behavior. Extensive experiments on a large real-world dataset demonstrate that our method outperforms classical models and ViT-based radar chart baselines, yielding gains of 17.7 in F1 score, 29.4 in precision, and 16.1 in AUC, along with improved interpretability. The framework's modular design, explainability tools, and efficient deployment characteristics make it suitable for large-scale churn modeling in dynamic gig-economy platforms.

Figures

Figures reproduced from arXiv: 2506.17325 by the authors.

Figure 1
Figure 1. Examples of radar chart images representing daily courier behavior. Each [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Overview of our hybrid CNN+LSTM churn prediction framework. Daily [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 4
Figure 4. UMAP of sequence embeddings. Blue: retained users; red: churners. Overlap highlights behavioral com￾plexity and supports temporal over static modeling. Figures 3 and 5 further validate the robustness of our model. The static ma￾trix confirms balanced classification with minimal error, while the temporal plot shows that performance stabilizes within the first few epochs. The low variance in false predictions over tim… view at source ↗
Figures from the paper (2 more)
Figure 5
Figure 5. Figure 5: Temporal evolution of confusion matrix components over 20 training [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]
Figure 6
Figure 6. Figure 6: Validation and test accuracy across epochs: curves stay closely aligned, [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

101 extracted references · 71 canonical work pages

  1. [58]

    Physical review letters89(1), 015002 (2002) 16 S

    Mozer, F., Bale, S., Phan, T.: Evidence of diffusion regions at a subsolar magne- topause crossing. Physical review letters89(1), 015002 (2002) 16 S. Najafi et al

    work page 2002

  2. [1]

    Journal of Computer Science and Technology Studies6(4), 92–101 (2024)

    Ahmed, M.P., Arif, M., Al Mamun, A., Mahmud, F., Rahman, T., Ahmmed, M.J., Mou, S.N., Akter, P., Chowdhury, M.S.R., Uddin, M.K.: A comparative study of machine learning models for predicting customer churn in retail bank- ing: Insights from logistic regression, random forest, gbm, and svm. Journal of Computer Science and Technology Studies6(4), 92–101 (2024)

    2024

  3. [2]

    Available at SSRN 3969455 (2021) Title Suppressed Due to Excessive Length 13

    Ahn, D., Lee, D., Hosanagar, K.: Modeling lengthy behavioral log data for cus- tomer churn management: A representation learning approach. Available at SSRN 3969455 (2021) Title Suppressed Due to Excessive Length 13

    2021

  4. [3]

    International Journal of Research in Humanities and Social Studies10(1), 31–42 (2023)

    Al Salmi, A.Z., Khalifa, R.Z.: Analyzing the advertising language of talabat & hunger station food delivery apps using grice maxims of cooperative principles. International Journal of Research in Humanities and Social Studies10(1), 31–42 (2023)

    2023

  5. [4]

    In: 18 International Multidisciplinary Scientific Geoconference SGEM 2018: Conference Proceedings

    Aleksandrova, Y.: Application of machine learning for churn prediction based on transactional data (rfm analysis). In: 18 International Multidisciplinary Scientific Geoconference SGEM 2018: Conference Proceedings. vol. 18, pp. 125–132 (2018)

    2018

  6. [5]

    In: International Conference on Big Data and Smart Digital Environment

    Al_Janabi, S., Razaq, F.: Intelligent big data analysis to design smart predictor for customer churn in telecommunication industry. In: International Conference on Big Data and Smart Digital Environment. pp. 246–272. Springer (2018)

    2018

  7. [6]

    AMAR (An- dalas Management Review)5(2), 95–123 (2021)

    Amsal, A.A., Albar, B.B., Yeni, Y.H.: Systematic literature review on customer switching behaviour from marketing and data science perspectives. AMAR (An- dalas Management Review)5(2), 95–123 (2021)

    2021

  8. [7]

    Andersson, V.: Developing a systematic framework for churn mitigation in a b2b saas environment (2024)

    2024

Show all 101 references
  1. [8]

    Benner, C., Mason, S., Carré, F., Tilly, C.: Delivering insecurity: E-commerce and the future of work in food retail (2020)

    2020

  2. [9]

    Netherlands (2019)

    Bernat, J., Koning, A., Fok, D.: Modelling customer lifetime value in a continuous, non-contractual time setting. Netherlands (2019)

    2019

  3. [10]

    Blattberg, R.C., Kim, B.D., Neslin, S.A.: Database marketing: Analyzing and managing customers (2008)

    2008

  4. [11]

    Journal of Interactive Marketing 23(2), 157–168 (2009)

    Blattberg,R.C.,Malthouse,E.C.,Neslin,S.A.:Customerlifetimevalue:Empirical generalizations and some conceptual questions. Journal of Interactive Marketing 23(2), 157–168 (2009)

    2009

  5. [12]

    Mathematics11(5), 1137 (2023)

    Bogaert, M., Delaere, L.: Ensemble methods in customer churn prediction: a com- parative analysis of the state-of-the-art. Mathematics11(5), 1137 (2023)

    2023

  6. [13]

    European journal of operational research164(1), 252–268 (2005)

    Buckinx, W., Van den Poel, D.: Customer base analysis: partial defection of be- haviourally loyal clients in a non-contractual fmcg retail setting. European journal of operational research164(1), 252–268 (2005)

    2005

  7. [14]

    Expert Systems with Applications36(3), 4626–4636 (2009)

    Burez, J., Van den Poel, D.: Handling class imbalance in customer churn predic- tion. Expert Systems with Applications36(3), 4626–4636 (2009)

    2009

  8. [15]

    Algorithms17(6), 231 (2024)

    Chang, V., Hall, K., Xu, Q.A., Amao, F.O., Ganatra, M.A., Benson, V.: Predic- tion of customer churn behavior in the telecommunication industry using machine learning models. Algorithms17(6), 231 (2024)

    2024

  9. [16]

    Information Systems and e- Business Management13, 475–494 (2015)

    Chen, K., Hu, Y.H., Hsieh, Y.C.: Predicting customer churn from valuable b2b customers in the logistics industry: a case study. Information Systems and e- Business Management13, 475–494 (2015)

    2015

  10. [17]

    arXiv preprint arXiv:2406.07476 (2024)

    Cheng, Z., Leng, S., Zhang, H., Xin, Y., Li, X., Chen, G., Zhu, Y., Zhang, W., Luo, Z., Zhao, D., et al.: Videollama 2: Advancing spatial-temporal modeling and audio understanding in video-llms. arXiv preprint arXiv:2406.07476 (2024)

    2024 arXiv

  11. [18]

    In: International Conference on Information Integration and Web In- telligence

    Coolwijk, S., Ziabari, S.S.M., Angileri, F.: Vision transformer approach to cus- tomer churn prediction radar chart image classification for non-subscription based e-commerce. In: International Conference on Information Integration and Web In- telligence. pp. 75–80. Springer (2025)

    2025

  12. [19]

    Information Process- ing & Management (2021)

    Dai, X., Wang, P.: Cost optimization in retention strategies. Information Process- ing & Management (2021)

    2021

  13. [20]

    Journal of Discrete Mathematical Sciences and Cryptography25(7), 1965–1985 (2022) 14 S

    De, S., Prabu, P.: Predicting customer churn: A systematic literature review. Journal of Discrete Mathematical Sciences and Cryptography25(7), 1965–1985 (2022) 14 S. Najafi et al

    2022

  14. [21]

    IEEE Transactions on Geoscience and Remote Sensing (2024)

    Ding, L., Zhang, J., Guo, H., Zhang, K., Liu, B., Bruzzone, L.: Joint spatio- temporal modeling for semantic change detection in remote sensing images. IEEE Transactions on Geoscience and Remote Sensing (2024)

    2024

  15. [22]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Donahue, J., Anne Hendricks, L., Guadarrama, S., Rohrbach, M., Venugopalan, S., Saenko, K., Darrell, T.: Long-term recurrent convolutional networks for visual recognition and description. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 26...

    2015

  16. [23]

    Journal of Labor Economics (2023)

    Dong, F., Lim, Y.: Worker satisfaction and retention in platform delivery. Journal of Labor Economics (2023)

    2023

  17. [24]

    Manufacturing & Service Operations Management22(1), 191–202 (2020)

    Donohue, K., Özer, Ö., Zheng, Y.: Behavioral operations: Past, present, and fu- ture. Manufacturing & Service Operations Management22(1), 191–202 (2020)

    2020

  18. [25]

    an image isworth 16×16 words: transformers for im- age recognition atscale

    Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X.: Thomas un- terthiner mostafa dehghani matthias minderer georg heigold sylvain gelly jakob uszkoreit and neil houlsby. an image isworth 16×16 words: transformers for im- age recognition atscale. In: Internat...

    2021

  19. [26]

    arXiv preprint arXiv:2010.11929 (2020)

    Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., et al.: An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929 (2020)

    2020 arXiv

  20. [27]

    Journal of Consumer Satisfaction, Dissatisfaction and Complaining Behavior 14, 46–54 (2001)

    East, R., Lomax, W., Narain, R.: Customer tenure, recommendation and switch- ing. Journal of Consumer Satisfaction, Dissatisfaction and Complaining Behavior 14, 46–54 (2001)

    2001

  21. [28]

    Marketing Science29(6), 1086–1108 (2010)

    Fader, P.S., Hardie, B.G., Shang, J.: Customer-base analysis in a discrete-time noncontractual setting. Marketing Science29(6), 1086–1108 (2010)

    2010

  22. [29]

    In: Proceedings of the 29th ACM International Conference on Multimedia

    Fu, Z., Liu, F., Zhang, J., Wang, H., Yang, C., Xu, Q., Qi, J., Fu, X., Zhou, A.: Sagn: Semantic adaptive graph network for skeleton-based human action recogni- tion. In: Proceedings of the 29th ACM International Conference on Multimedia. pp. 197–205 (2021). https://doi.org/10...

    2021

  23. [30]

    In: Practical Strategies and Case Studies for Online Marketing 6.0, pp

    Ganapathy, V.: Decoding customer engagement: A data-driven analysis of food delivery apps in bangalore. In: Practical Strategies and Case Studies for Online Marketing 6.0, pp. 251–298. IGI Global Scientific Publishing (2025)

    2025

  24. [31]

    Data Mining and Knowledge Discovery (2023)

    Gao, L., Tan, Z.: Generalizing churn predictors across regional delivery markets. Data Mining and Knowledge Discovery (2023)

    2023

  25. [32]

    Industrial Marketing Management107, 134–147 (2022)

    Gattermann-Itschert, T., Thonemann, U.W.: Proactive customer retention man- agement in a non-contractual b2b setting based on churn prediction with random forests. Industrial Marketing Management107, 134–147 (2022)

    2022

  26. [33]

    Granov,A.:Customerloyalty,returnandchurnpredictionthroughmachinelearn- ing methods: for a swedish fashion and e-commerce company (2021)

    2021

  27. [34]

    IEEE Internet of Things Journal (2024)

    Guo, S., Deng, B., Chen, C., Ke, J., Wang, J., Long, S., Xu, K.: Seeking in ride-on- demand service: A reinforcement learning model with dynamic price prediction. IEEE Internet of Things Journal (2024)

    2024

  28. [35]

    Gupta,S.:Researchnote—channelstructurewithknowledgespillovers.Marketing Science27(2), 247–261 (2008)

    2008

  29. [36]

    Computers & Operations Re- search34(10), 2902–2917 (2007)

    Hadden, J., Tiwari, A., Roy, R., Ruta, D.: Computer assisted customer churn management: State-of-the-art and future trends. Computers & Operations Re- search34(10), 2902–2917 (2007)

    2007

  30. [37]

    Ilr Review71(3), 705–732 (2018)

    Hall, J.V., Krueger, A.B.: An analysis of the labor market for uber’s driver- partners in the united states. Ilr Review71(3), 705–732 (2018)

    2018

  31. [38]

    Greenhaven Publishing LLC (2020) Title Suppressed Due to Excessive Length 15

    Heing, B.: The Gig Economy. Greenhaven Publishing LLC (2020) Title Suppressed Due to Excessive Length 15

    2020

  32. [39]

    Neural computation 9(8), 1735–1780 (1997)

    Hochreiter, S., Schmidhuber, J.: Long short-term memory. Neural computation 9(8), 1735–1780 (1997)

    1997

  33. [40]

    Expert Systems with Applications39(1), 1414–1425 (2012)

    Huang, B., Kechadi, M.T., Buckley, B.: Customer churn prediction in telecom- munications. Expert Systems with Applications39(1), 1414–1425 (2012)

    2012

  34. [41]

    Computers & Electrical Engineering38(6), 1808–1819 (2012)

    Idris, A., Rizwan, M., Khan, A.: Churn prediction in telecom using random for- est and pso based data balancing in combination with various feature selection strategies. Computers & Electrical Engineering38(6), 1808–1819 (2012)

    2012

  35. [42]

    Imani, M.: Customer churn prediction in telecommunication industry: A (2024)

    2024

  36. [43]

    Statistical Analysis and Data Mining: The ASA Data Science Journal13(3), 245–260 (2020)

    Jastrzebska, A.: Lagged encoding for image-based time series classification using convolutional neural networks. Statistical Analysis and Data Mining: The ASA Data Science Journal13(3), 245–260 (2020)

    2020

  37. [44]

    International Journal of Research in All Subjects in Multi Languages13(1), 153 (2025)

    Katyayan, S.S., Singh, S.: Optimizing consumer retention strategies through data- driven insights in digital marketplaces. International Journal of Research in All Subjects in Multi Languages13(1), 153 (2025)

    2025

  38. [45]

    Scientific Reports13(1), 17294 (2023)

    Khattak, A., Mehak, Z., Ahmad, H., Asghar, M.U., Asghar, M.Z., Khan, A.: Customer churn prediction using composite deep learning technique. Scientific Reports13(1), 17294 (2023)

    2023

  39. [46]

    Journal of Systems and Information Technology19(1/2), 65–93 (2017)

    Khodabandehlou, S., Zivari Rahman, M.: Comparison of supervised machine learning techniques for customer churn prediction based on analysis of customer behavior. Journal of Systems and Information Technology19(1/2), 65–93 (2017)

    2017

  40. [47]

    IEEE Transactions on Artificial Intelligence (2025)

    Kougioumtzidis, G., Poulkov, V.K., Lazaridis, P.I., Zaharis, Z.D.: Mobile net- work traffic prediction using temporal fusion transformer. IEEE Transactions on Artificial Intelligence (2025)

    2025

  41. [48]

    Turkish Online Journal of Qualita- tive Inquiry12(3) (2021)

    Krishnan, K.P., Kombaiya, A.K.: Customer churn prediction in telecom industry using dwhbi approaches and r programming. Turkish Online Journal of Qualita- tive Inquiry12(3) (2021)

    2021

  42. [49]

    Advances in Neural Information Processing Systems36, 49187–49204 (2023)

    Li, Z., Li, S., Yan, X.: Time series as images: Vision transformer for irregularly sampled time series. Advances in Neural Information Processing Systems36, 49187–49204 (2023)

    2023

  43. [50]

    Applied Intelligence (2024)

    Liu, S., Wang, Y., Zhao, H.: Hybrid bilstm-cnn for churn prediction. Applied Intelligence (2024)

    2024

  44. [51]

    Journal of Organizational and End User Computing (JOEUC)36(1), 1–23 (2024)

    Liu, Y., Hou, J., Zhao, W.: Deep learning and user consumption trends classifi- cation and analysis based on shopping behavior. Journal of Organizational and End User Computing (JOEUC)36(1), 1–23 (2024)

    2024

  45. [52]

    IEEE transactions on information theory28(2), 129–137 (1982)

    Lloyd, S.: Least squares quantization in pcm. IEEE transactions on information theory28(2), 129–137 (1982)

    1982

  46. [53]

    Production and Operations Management30(6), 1927–1943 (2021)

    Lu, M., Shen, Z.J.M.: A review of robust operations management under model uncertainty. Production and Operations Management30(6), 1927–1943 (2021)

    2021

  47. [54]

    Advances in Neural Information Processing Systems30(2017)

    Lundberg, S.M., Lee, S.I.: A unified approach to interpreting model predictions. Advances in Neural Information Processing Systems30(2017)

    2017

  48. [55]

    Journal of Theoretical and Applied Electronic Commerce Research17(1), 165–198 (2022)

    Matuszelański, K., Kopczewska, K.: Customer churn in retail e-commerce busi- ness: Spatial and machine learning approach. Journal of Theoretical and Applied Electronic Commerce Research17(1), 165–198 (2022)

    2022

  49. [56]

    Annals of Operations Research339(1), 765–787 (2024)

    Mena, G., Coussement, K., De Bock, K.W., De Caigny, A., Lessmann, S.: Exploit- ing time-varying rfm measures for customer churn prediction with deep neural networks. Annals of Operations Research339(1), 765–787 (2024)

    2024

  50. [57]

    arXiv preprint arXiv:1710.03740 (2017)

    Micikevicius, P., Narang, S., Alben, J., Diamos, G., Elsen, E., Garcia, D., Gins- burg, B., Houston, M., Kuchaiev, O., Venkatesh, G., et al.: Mixed precision train- ing. arXiv preprint arXiv:1710.03740 (2017)

    2017 arXiv

  51. [59]

    International Journal of Developing and Emerging Economies8(1), 1–16 (2020)

    Mukhopadhyay, B.R., Chatwin, C.: Your driver is didi and minutes away from your pick-up point’: understanding employee motivation in the gig economy of china. International Journal of Developing and Emerging Economies8(1), 1–16 (2020)

    2020

  52. [60]

    Muñoz, L.E., et al.: Customer churn detection and marketing retention strategies in the online food delivery business (2022)

    2022

  53. [61]

    https://github.com/uploadcare/pillow-simd (2018)

    Murray, A.C.: pillow-simd. https://github.com/uploadcare/pillow-simd (2018)

    2018

  54. [62]

    Journal of service research9(2), 95–112 (2006)

    Neslin, S.A., Grewal, D., Leghorn, R., Shankar, V., Teerling, M.L., Thomas, J.S., Verhoef, P.C.: Challenges and opportunities in multichannel customer manage- ment. Journal of service research9(2), 95–112 (2006)

    2006

  55. [63]

    Oblander, E.S., McCarthy, D.: How has covid-19 impacted customer relationship dynamics at restaurant food delivery businesses. Mark. Sci. Inst. Work. Pap. Ser 23(2139), 35 (2021)

    2021

  56. [64]

    Osbat, C., Conflitti, C., Bellocca, G.P., Eiglsperger, M., Goldhammer, B., Kuik, F.,Menz,J.O.,Rumler,F.,SaezMoreno,M.,Segers,L.,etal.:Measuringinflation with heterogeneous preferences, taste shifts and product innovation: methodolog- ical challenges and evidence from microdata...

    2023

  57. [65]

    Com- puters & Industrial Engineering193, 110318 (2024)

    Paiva, R.G., Melo, Y.R., Cavalcante, C.A., Tenório, V.A., Do, P.: Developing data-driven o&m policy through sequential pattern mining: A case study. Com- puters & Industrial Engineering193, 110318 (2024)

    2024

  58. [66]

    Journal of Current Science and Technology13(1), 136–161 (2023)

    Panimalar, S.A., Krishnakumar, A.: A review of churn prediction models using different machine learning and deep learning approaches in cloud environment. Journal of Current Science and Technology13(1), 136–161 (2023)

    2023

  59. [67]

    In: The International Conference on Artificial Intel- ligence and Computer Vision

    Rabbah, J., Ridouani, M., Hassouni, L.: New approach to telecom churn predic- tion based on transformers. In: The International Conference on Artificial Intel- ligence and Computer Vision. pp. 565–574. Springer (2023)

    2023

  60. [68]

    Harvard business review68(5), 105–111 (1990)

    Reichheld, F.F., Sasser, W.E.: Zero defeofions: Quoliiy comes to services. Harvard business review68(5), 105–111 (1990)

    1990

  61. [69]

    IEEE transactions on visualization and computer graphics 23(1), 241–250 (2016)

    Sacha, D., Zhang, L., Sedlmair, M., Lee, J.A., Peltonen, J., Weiskopf, D., North, S.C., Keim, D.A.: Visual interaction with dimensionality reduction: A structured literature analysis. IEEE transactions on visualization and computer graphics 23(1), 241–250 (2016)

    2016

  62. [70]

    In: CVPR (2018)

    Sandler, M., Howard, A., Zhu, M., Zhmoginov, A., Chen, L.C.: Mobilenetv2: Inverted residuals and linear bottlenecks. In: CVPR (2018)

    2018

  63. [71]

    IEEE International Conference on Computer Vision pp

    Selvaraju, R.R., et al.: Grad-cam: Visual explanations from deep networks. IEEE International Conference on Computer Vision pp. 618–626 (2017)

    2017

  64. [72]

    arXiv preprint arXiv:2501.18875 (2025)

    Sepanj, M.H., Ghojogh, B., Fieguth, P.: Self-supervised learning using nonlinear dependence. arXiv preprint arXiv:2501.18875 (2025)

    2025

  65. [73]

    Big Data and Cognitive Computing9(4), 84 (2025)

    Sepanj, M.H., Moradi, S., Azimifar, Z., Fieguth, P.: Uncertainty-awareδ-glmb filtering for multi-target tracking. Big Data and Cognitive Computing9(4), 84 (2025)

    2025

  66. [74]

    European Advanced AI Journal11(8) (2022)

    Sharma, A., Patel, N., Gupta, R.: Enhancing predictive customer retention using machine learning algorithms: A comparative study of random forest, xgboost, and neural networks. European Advanced AI Journal11(8) (2022)

    2022

  67. [75]

    arXiv preprint arXiv:1807.08207 (2018)

    Sheil, H., Rana, O., Reilly, R.: Predicting purchasing intent: automatic feature learning using recurrent neural networks. arXiv preprint arXiv:1807.08207 (2018)

    2018 arXiv

  68. [76]

    Journal of Industrial Economics (2022) Title Suppressed Due to Excessive Length 17

    Silverman, B., Long, T.: The economic sustainability of gig economy platforms. Journal of Industrial Economics (2022) Title Suppressed Due to Excessive Length 17

    2022

  69. [77]

    Discrete Dynamics in Nature and Society2022(1), 5134356 (2022)

    Sina Mirabdolbaghi, S.M., Amiri, B.: Model optimization analysis of customer churn prediction using machine learning algorithms with focus on feature reduc- tions. Discrete Dynamics in Nature and Society2022(1), 5134356 (2022)

    2022

  70. [78]

    In: International conference on machine learning

    Srivastava, N., Mansimov, E., Salakhudinov, R.: Unsupervised learning of video representations using lstms. In: International conference on machine learning. pp. 843–852. PMLR (2015)

    2015

  71. [79]

    Information Systems Research35(3), 1216–1234 (2024)

    Sun, G., Kim, Y., Tan, Y., Parker, G.G.: Dinner at your doorstep: Service in- novation via the gig economy on food delivery platforms. Information Systems Research35(3), 1216–1234 (2024)

    2024

  72. [80]

    In: 2017 International Joint Conference on Neural Networks (IJCNN)

    Sung, C., Higgins, C.Y., Zhang, B., Choe, Y.: Evaluating deep learning in chum prediction for everything-as-a-service in the cloud. In: 2017 International Joint Conference on Neural Networks (IJCNN). pp. 3664–3669. IEEE (2017)

    2017

  73. [81]

    Advances in Neural Information Processing Systems35, 38204–38217 (2022)

    Tatsunami, Y., Taki, M.: Sequencer: Deep lstm for image classification. Advances in Neural Information Processing Systems35, 38204–38217 (2022)

    2022

  74. [82]

    University of Antwerp (2019)

    Torfs, E.: In vitro biological investigation of novel anti-tubercular compound classes and the development of improved research tools. University of Antwerp (2019)

    2019

  75. [83]

    IEEE access 6, 1155–1166 (2017)

    Ullah, A., Ahmad, J., Muhammad, K., Sajjad, M., Baik, S.W.: Action recognition in video sequences using deep bi-directional lstm with cnn features. IEEE access 6, 1155–1166 (2017)

    2017

  76. [84]

    Simulation Modelling Practice and Theory55, 1–9 (2015)

    Vafeiadis, T., Diamantaras, K.I., Sarigiannidis, G., Chatzisavvas, K.C.: A com- parison of machine learning techniques for customer churn prediction. Simulation Modelling Practice and Theory55, 1–9 (2015)

    2015

  77. [85]

    Advances in Neural Information Processing Systems30(2017)

    Vaswani, A., et al.: Attention is all you need. Advances in Neural Information Processing Systems30(2017)

    2017

  78. [86]

    IEEE transactions on knowledge and data engineering25(5), 961–973 (2012)

    Verbraken, T., Verbeke, W., Baesens, B.: A novel profit maximizing metric for measuring classification performance of customer churn prediction models. IEEE transactions on knowledge and data engineering25(5), 961–973 (2012)

    2012

  79. [87]

    Expert Systems with Applications248, 123352 (2024)

    Wang, C., Rao, C., Hu, F., Xiao, X., Goh, M.: Risk assessment of customer churn in telco using fclcnn-lstm model. Expert Systems with Applications248, 123352 (2024)

    2024

  80. [88]

    Manufacturing & Service Operations Management (2025)

    Wang, L., Webster, S., Rabinovich, E.: Structural estimation of attrition in a last-mile delivery platform: The role of driver heterogeneity, compensation, and experience. Manufacturing & Service Operations Management (2025)

    2025

  81. [89]

    arXiv preprint arXiv:1506.00327 (2015)

    Wang, Z., Oates, T.: Imaging time-series to improve classification and imputation. arXiv preprint arXiv:1506.00327 (2015)

    2015 arXiv

  82. [90]

    ACM Transactions on Knowledge Discovery from Data (2020)

    Wei, X., Zhou, Y.: Cross-platform churn modeling: Challenges and insights. ACM Transactions on Knowledge Discovery from Data (2020)

    2020

  83. [91]

    Science China Information Sciences 63, 1–21 (2020)

    Xie, G., Shangguan, A., Fei, R., Ji, W., Ma, W., Hei, X.: Motion trajectory pre- diction based on a cnn-lstm sequential model. Science China Information Sciences 63, 1–21 (2020)

    2020

  84. [92]

    arXiv preprint arXiv:2109.00724 (2021)

    Yang, L., Niu, X., Wu, J.: Rf-lighgbm: A probabilistic ensemble way to pre- dict customer repurchase behaviour in community e-commerce. arXiv preprint arXiv:2109.00724 (2021)

    2021 arXiv

  85. [93]

    IEEE Access (2022)

    Yang, W., Chen, X., Hu, B.: Cnn-lstm hybrid for telecom churn. IEEE Access (2022)

    2022

  86. [94]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition

    Yang, X., Molchanov, P., Kautz, J.: Making convolutional networks recurrent for visual sequence learning. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. pp. 6469–6478 (2018) 18 S. Najafi et al

    2018

  87. [95]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Yue-Hei Ng, J., Hausknecht, M., Vijayanarasimhan, S., Vinyals, O., Monga, R., Toderici, G.: Beyond short snippets: Deep networks for video classification. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 4694–4702 (2015)

    2015

  88. [96]

    Master’s thesis, University of Missouri-Columbia (2023)

    Zahin, M.A.: Multi-Headed Self-Attention Mechanism-Based Transformer Model for Predicting Bus Travel Times Across Multiple Bus Routes Using Heterogeneous Datasets. Master’s thesis, University of Missouri-Columbia (2023)

    2023

  89. [97]

    IEEE transactions on Big Data6(1), 3–28 (2018)

    Zhang, D., Yin, J., Zhu, X., Zhang, C.: Network representation learning: A survey. IEEE transactions on Big Data6(1), 3–28 (2018)

    2018

  90. [98]

    Expert Systems with Applications (2023)

    Zhang, L., Feng, D., Li, H.: Comparative study of deep models in customer attri- tion. Expert Systems with Applications (2023)

    2023

  91. [99]

    AI Open (2022)

    Zhang, T., Liu, Y.: Explainable ai in churn prediction using shap. AI Open (2022)

    2022

  92. [100]

    Zhang, X., Guo, F., Chen, T., Pan, L., Beliakov, G., Wu, J.: A brief survey of machinelearninganddeeplearningtechniquesfore-commerceresearch.Journalof Theoretical and Applied Electronic Commerce Research18(4), 2188–2216 (2023)

    2023

  93. [101]

    Marketing science34(2), 195–208 (2015)

    Zhang, Y., Bradlow, E.T., Small, D.S.: Predicting customer value using clumpi- ness: From rfm to rfmc. Marketing science34(2), 195–208 (2015)

    2015

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.