![Introduction to dynamic resolution encoding
2
[1] A. Aaron, Z. Li, M. Manohara, J.D. Cock, D. Ronca, "Per-title encode optimization." The Netflix Techblog (2015).
Motivation
● Traditional dynamic resolution encoding schemes are
based on the fact that one resolution performs better than
others in a scene for a given bitrate range.
● The performance (quality) of resolutions will differ from
one codec to the other.
● The more efficient the codec, the less bitrate it needs
to encode at a specific resolution with acceptable
quality.
Figure: Conceptual plot to depict the bitrate-quality
relationship for any video source encoded at different
resolutions. Source: [1]](https://image.slidesharecdn.com/videostreaming4amrithapremkumarrptu-240212171041-d94e267c/85/Energy-efficient-Adaptive-Video-Streaming-with-Latency-Aware-Dynamic-Resolution-Encoding-2-320.jpg)
![Introduction to dynamic resolution encoding
3
[2] G. Sullivan, J. Ohm, W. Han, and T. Wiegand, “Overview of the high efficiency video coding (HEVC) standard,” in IEEE Transactions on circuits and systems for video technology, vol. 22, no.
12. IEEE, 2012, pp. 1649–1668.
[3] B. Bross, Y. Wang, Y. Ye, S. Liu, J. Chen, G. Sullivan, and J. Ohm. (2021). Overview of the Versatile Video Coding (VVC) Standard and its Applications. IEEE Transactions on Circuits and
Systems for Video Technology. 31. 3736-3764. 10.1109/TCSVT.2021.3101953.
[4] A. Stergiou and R. Poppe, “AdaPool: Exponential Adaptive Pooling for Information-Retaining Downsampling,” in IEEE Transactions on Image Processing, vol. 32, 2023, pp. 251–266.
In the context of VVC…
● Versatile Video Coding (VVC) is the successor of High Efficiency Video Coding (HEVC) [2], with better
compression efficiency [3].
● The best resolution for a given bitrate depends on the video complexity.
Figure: Rate-distortion (RD) curves of representative sequences (segments) of Inter-4K [4] dataset encoded at 540p, 1080p, and 2160p
resolutions using VVenC at faster preset. Here, XPSNR is used as the quality metric.](https://image.slidesharecdn.com/videostreaming4amrithapremkumarrptu-240212171041-d94e267c/85/Energy-efficient-Adaptive-Video-Streaming-with-Latency-Aware-Dynamic-Resolution-Encoding-3-320.jpg)
![Introduction to dynamic resolution encoding
5
Encoding time-constraint- Green encoding?
Lower encoding latency aligns with eco-friendly streaming practices by optimizing resource
utilization and minimizing unnecessary energy expenditure.
● Reducing encoding energy consumption (in data centers) is critical in streaming applications
since it contributes to environmental sustainability [5].
● Some researches suggest a pseudo linear relationship between encoding time and encoding
energy consumption [6].
[5] A. Stephens, C. Tremlett-Williams, L. Fitzpatrick, L. Acerini, M. Anderson, and N. Crabbendam, “The Carbon Impacts of Video Streaming,” Jun. 2021.
[6] V. V. Menon, S. Afzal, P. T. Rajendran, K. Schoeffmann, R. Prodan, and C. Timmerer. [n. d.]. Content-Adaptive Variable Framerate Encoding Scheme for Green Live
Streaming, [Online]. Available: http://arxiv.org/abs/2311.08074](https://image.slidesharecdn.com/videostreaming4amrithapremkumarrptu-240212171041-d94e267c/85/Energy-efficient-Adaptive-Video-Streaming-with-Latency-Aware-Dynamic-Resolution-Encoding-5-320.jpg)
![Related work
6
Table: Comparison of the state-of-the-art dynamic resolution per-
title encoding methods with LADRE.
[3] J. Cock, Zhi Li, M. Manohara, and A. Aaron, “Complexity-based consistent-quality encoding in the cloud,” in 2016 IEEE International Conference on Image Processing (ICIP), 2016, pp.
1484–1488.
[4] A. Katsenou, J. Sole, and D. R. Bull, “Content-gnostic Bitrate Ladder Prediction for Adaptive Video Streaming,” in 2019 Picture Coding Symposium (PCS), 2019, pp. 1–5.
[5] A. Zabrovskiy, P. Agrawal, C. Timmerer, and R. Prodan, “FAUST: Fast Per-Scene Encoding Using Entropy-Based Scene Detection and Machine Learning,” in 2021 30th Conference of
Open Innovations Association FRUCT, 2021, pp. 292–302.
[6] M. Bhat, J. Thiesse, and P. Le Callet, “Combining Video Quality Metrics To Select Perceptually Accurate Resolution In A Wide Quality Range: A Case Study,” in 2021 IEEE
International Conference on Image Processing (ICIP), 2021, pp. 2164–2168.
[7] V. V. Menon, P. T. Rajendran, C. Feldmann, K. Schoeffmann, M. Ghanbari and C. Timmerer, "JND-Aware Two-Pass Per-Title Encoding Scheme for Adaptive Live Streaming," in IEEE
Transactions on Circuits and Systems for Video Technology, vol. 34, no. 2, pp. 1281-1294, Feb. 2024, doi: 10.1109/TCSVT.2023.3290725.
• Current related work lacks considering encoding
latency constraints while selecting the optimized
encoding resolution.
• Most state-of-the-art methods need pre-encodings
which yield significant latency and energy
consumption.
• As an example, for bruteforce, the total time taken is
the total time of all encodings and the quality
evaluation of those encodings.](https://image.slidesharecdn.com/videostreaming4amrithapremkumarrptu-240212171041-d94e267c/85/Energy-efficient-Adaptive-Video-Streaming-with-Latency-Aware-Dynamic-Resolution-Encoding-6-320.jpg)
![Spatiotemporal complexity feature extraction
8
We use seven DCT-energy-based features extracted using Video Complexity Analyzer (VCA)
[8]:
● average texture energy (EY),
● average gradient of the luma texture energy (h)
● average luma brightness (LY),
● average chroma texture energy of U and V channels (EU and EV)
● average chroma brightness of U and V channels (LU and LV) [8].
[8] V. V. Menon, C. Feldmann, K. Schoeffmann, M. Ghanbari, and C. Timmerer, “Green Video Complexity Analysis for Efficient Encoding in Adaptive Video Streaming,” in First
International ACM Green Multimedia Systems Workshop (GMSys ’23), 2023.](https://image.slidesharecdn.com/videostreaming4amrithapremkumarrptu-240212171041-d94e267c/85/Energy-efficient-Adaptive-Video-Streaming-with-Latency-Aware-Dynamic-Resolution-Encoding-8-320.jpg)
![Optimized resolution prediction
9
[9] V. V. Menon et al. Content-Adaptive Variable Framerate Encoding Scheme for Green Live Streaming. 2023. arXiv: 2311.08074 [cs.MM]. url: https://arxiv.org/pdf/2311.08074.
[10] V. V. Menon et al. “Energy-Efficient Multi-Codec Bitrate Ladder Estimation for Adaptive Video Streaming”. In: 2023 International Conference on Visual Communications and Image
Processing (VCIP). 2023..
Modeling
• The perceptual quality and encoding time of the representation (rt , bt ) rely on the
extracted video complexity features, encoding resolution, and target bitrate [9, 10]:
• A higher resolution, and/or bitrate may improve the quality and increase the file size of the
encoded video segment.
• Similarly, a higher resolution, and/or bitrate can increase the encoding duration.](https://image.slidesharecdn.com/videostreaming4amrithapremkumarrptu-240212171041-d94e267c/85/Energy-efficient-Adaptive-Video-Streaming-with-Latency-Aware-Dynamic-Resolution-Encoding-9-320.jpg)
![Optimized resolution prediction
10
[11] C. R. Helmrich et al., “Information on and analysis of the VVC encoders in the SDR UHD verification test,” in WG 05 MPEG Joint Video Coding Team(s) with ITU-T SG 16,
document JVET-T0103, Oct. 2020.
[12] M. Wien and V. Baroncini, “Report on VVC compression performance verification testing in the SDR UHD Random Access Category,” in WG 05 MPEG Joint Video Coding
Team(s) with ITU-T SG 16, document JVET-T0097, Oct. 2020.
[13] C. R. Helmrich et al., “A study of the extended perceptually weighted peak signal-to-noise ratio (XPSNR) for video compression with different resolutions and bit depths,” in ITU
Journal: ICT Discoveries, vol. 3, May 2020.
Optimization
• LADRE optimizes the perceptual quality (in terms of XPSNR) of encoded video
segments while adhering to real-time processing constraints. The optimization function
is:
We consider XPSNR as the perceptual quality measure instead of the popular VMAF.
• The correlation of XPSNR with subjective quality scores is higher than VMAF for
VVC-coded UHD bitstreams [11,12,13].](https://image.slidesharecdn.com/videostreaming4amrithapremkumarrptu-240212171041-d94e267c/85/Energy-efficient-Adaptive-Video-Streaming-with-Latency-Aware-Dynamic-Resolution-Encoding-10-320.jpg)
![Optimized rate factor prediction
11
[14] V. V Menon, H. Amirpour, M. Ghanbari, and C. Timmerer, “ETPS: Efficient Two-Pass Encoding Scheme for Adaptive Live Streaming,” in Proc. IEEE Int. Conf. Image Process.
(ICIP), Bordeaux, Oct. 2022.
Modeling
• Predicting the rate factor helps ensure consistent video quality throughout the stream.
• It allows the encoder to allocate bits judiciously, preventing under-allocation (resulting
in poor quality) or over-allocation (wasting bandwidth) of bits for encoding [14].
• The mathematical formulation of the rate factor optimization to yield a bitrate as close to
the target bitrate as possible can be expressed as follows:.
Optimization](https://image.slidesharecdn.com/videostreaming4amrithapremkumarrptu-240212171041-d94e267c/85/Energy-efficient-Adaptive-Video-Streaming-with-Latency-Aware-Dynamic-Resolution-Encoding-11-320.jpg)
![Experimental design
12
Experimental parameters
Table: Experimental parameters used to evaluate LADRE.
• The Inter-4K dataset [15] is used to validate the performance.
• The sequences are encoded using VVenC v1.10 [16] using preset 0 (faster).
• The spatiotemporal features are extracted using VCA v2.0.
• LADRE uses random forest regression models [17] to predict XPSNR, encoding time, and rate factor
trained for each supported resolution.
• Viewing resolution of UHD is assumed.
[15] A. Stergiou and R. Poppe, “AdaPool: Exponential Adaptive Pooling for Information-Retaining Downsampling,” in IEEE Transactions on Image Processing, vol. 32, 2023, pp. 251–266.
[16] A. Wieckowski, et. al., “VVenC: An Open And Optimized VVC Encoder Implementation,” in Proc. IEEE International Conference on Multimedia Expo Workshops (ICMEW), pp. 1–2.
[17] L. Breiman, “Random Forests,” Machine Learning, vol. 45, 2001.](https://image.slidesharecdn.com/videostreaming4amrithapremkumarrptu-240212171041-d94e267c/85/Energy-efficient-Adaptive-Video-Streaming-with-Latency-Aware-Dynamic-Resolution-Encoding-12-320.jpg)
![Experimental design
13
Benchmark schemes
Table: An example fixed bitrate ladder, i.e., set of
bitrate-resolution pairs. Source: [18].
1. Default: This method employs a fixed set of bitrate-
resolution pairs. We use the HLS bitrate ladder
specified in the Apple authoring specifications [18] as
the fixed set of bitrate-resolution pairs.
2. OPTE: This scheme predicts optimized resolution,
which yields the highest XPSNR for a given target
bitrate.
[18] Apple Inc., “HLS Authoring Specification for Apple Devices.” [Online]. Available: https://developer.apple.com/documentation/http-live-streaming/hls-authoring-specification-for-apple-
devices](https://image.slidesharecdn.com/videostreaming4amrithapremkumarrptu-240212171041-d94e267c/85/Energy-efficient-Adaptive-Video-Streaming-with-Latency-Aware-Dynamic-Resolution-Encoding-13-320.jpg)
![Experimental results
15
Table: Average results of the encoding schemes compared to the Default encoding.
● Coding efficiency (in terms of Bjøntegaard Delta [19] rates), and encoding times decrease as the
encoding time constraint is decreased.
● OPTE yields the highest coding efficiency (maximizes XPSNR with no encoding time constraint).
● We yield storage reduction using LADRE.
● Some high bitrate representations are eliminated because it cannot be encoded with the target
maximum encoding time.
● The maximum encoding time for a video segment in parallel encoding environment is shown in the
table below (last column). It is below the desired threshold.
[19] HSTP-VID-WPOM, “Working practices using objective metrics for evaluation of video coding efficiency experiments,” International Telecommunication Union, 2020. [Online].
Available: http://handle.itu.int/11.1002/pub/8160e8da-en](https://image.slidesharecdn.com/videostreaming4amrithapremkumarrptu-240212171041-d94e267c/85/Energy-efficient-Adaptive-Video-Streaming-with-Latency-Aware-Dynamic-Resolution-Encoding-15-320.jpg)
Energy-efficient Adaptive Video Streaming with Latency-Aware Dynamic Resolution Encoding
- 2. Introduction to dynamic resolution encoding 2 [1] A. Aaron, Z. Li, M. Manohara, J.D. Cock, D. Ronca, "Per-title encode optimization." The Netflix Techblog (2015). Motivation ● Traditional dynamic resolution encoding schemes are based on the fact that one resolution performs better than others in a scene for a given bitrate range. ● The performance (quality) of resolutions will differ from one codec to the other. ● The more efficient the codec, the less bitrate it needs to encode at a specific resolution with acceptable quality. Figure: Conceptual plot to depict the bitrate-quality relationship for any video source encoded at different resolutions. Source: [1]
- 3. Introduction to dynamic resolution encoding 3 [2] G. Sullivan, J. Ohm, W. Han, and T. Wiegand, “Overview of the high efficiency video coding (HEVC) standard,” in IEEE Transactions on circuits and systems for video technology, vol. 22, no. 12. IEEE, 2012, pp. 1649–1668. [3] B. Bross, Y. Wang, Y. Ye, S. Liu, J. Chen, G. Sullivan, and J. Ohm. (2021). Overview of the Versatile Video Coding (VVC) Standard and its Applications. IEEE Transactions on Circuits and Systems for Video Technology. 31. 3736-3764. 10.1109/TCSVT.2021.3101953. [4] A. Stergiou and R. Poppe, “AdaPool: Exponential Adaptive Pooling for Information-Retaining Downsampling,” in IEEE Transactions on Image Processing, vol. 32, 2023, pp. 251–266. In the context of VVC… ● Versatile Video Coding (VVC) is the successor of High Efficiency Video Coding (HEVC) [2], with better compression efficiency [3]. ● The best resolution for a given bitrate depends on the video complexity. Figure: Rate-distortion (RD) curves of representative sequences (segments) of Inter-4K [4] dataset encoded at 540p, 1080p, and 2160p resolutions using VVenC at faster preset. Here, XPSNR is used as the quality metric.
- 4. Introduction to dynamic resolution encoding 4 Encoding time ● Encoding time increases as the target encoding resolution increases owing to the increased computational complexity. Figure: Rate-encoding time curves of representative sequences (segments) of Inter-4K dataset encoded at 540p, 1080p, and 2160p resolutions using VVenC at faster preset.
- 5. Introduction to dynamic resolution encoding 5 Encoding time-constraint- Green encoding? Lower encoding latency aligns with eco-friendly streaming practices by optimizing resource utilization and minimizing unnecessary energy expenditure. ● Reducing encoding energy consumption (in data centers) is critical in streaming applications since it contributes to environmental sustainability [5]. ● Some researches suggest a pseudo linear relationship between encoding time and encoding energy consumption [6]. [5] A. Stephens, C. Tremlett-Williams, L. Fitzpatrick, L. Acerini, M. Anderson, and N. Crabbendam, “The Carbon Impacts of Video Streaming,” Jun. 2021. [6] V. V. Menon, S. Afzal, P. T. Rajendran, K. Schoeffmann, R. Prodan, and C. Timmerer. [n. d.]. Content-Adaptive Variable Framerate Encoding Scheme for Green Live Streaming, [Online]. Available: http://arxiv.org/abs/2311.08074
- 6. Related work 6 Table: Comparison of the state-of-the-art dynamic resolution per- title encoding methods with LADRE. [3] J. Cock, Zhi Li, M. Manohara, and A. Aaron, “Complexity-based consistent-quality encoding in the cloud,” in 2016 IEEE International Conference on Image Processing (ICIP), 2016, pp. 1484–1488. [4] A. Katsenou, J. Sole, and D. R. Bull, “Content-gnostic Bitrate Ladder Prediction for Adaptive Video Streaming,” in 2019 Picture Coding Symposium (PCS), 2019, pp. 1–5. [5] A. Zabrovskiy, P. Agrawal, C. Timmerer, and R. Prodan, “FAUST: Fast Per-Scene Encoding Using Entropy-Based Scene Detection and Machine Learning,” in 2021 30th Conference of Open Innovations Association FRUCT, 2021, pp. 292–302. [6] M. Bhat, J. Thiesse, and P. Le Callet, “Combining Video Quality Metrics To Select Perceptually Accurate Resolution In A Wide Quality Range: A Case Study,” in 2021 IEEE International Conference on Image Processing (ICIP), 2021, pp. 2164–2168. [7] V. V. Menon, P. T. Rajendran, C. Feldmann, K. Schoeffmann, M. Ghanbari and C. Timmerer, "JND-Aware Two-Pass Per-Title Encoding Scheme for Adaptive Live Streaming," in IEEE Transactions on Circuits and Systems for Video Technology, vol. 34, no. 2, pp. 1281-1294, Feb. 2024, doi: 10.1109/TCSVT.2023.3290725. • Current related work lacks considering encoding latency constraints while selecting the optimized encoding resolution. • Most state-of-the-art methods need pre-encodings which yield significant latency and energy consumption. • As an example, for bruteforce, the total time taken is the total time of all encodings and the quality evaluation of those encodings.
- 7. Latency-Aware Dynamic Resolution Encoding (LADRE) 7 Figure: Encoding using LADRE envisioned for video streaming. • Spatiotemporal complexity feature extraction, • Optimized resolution prediction, • Optimized rate factor prediction, and • Constrained variable bitrate (cVBR) encoding using the selected bitrate-resolution-rate factor combinations. Proposed architecture
- 8. Spatiotemporal complexity feature extraction 8 We use seven DCT-energy-based features extracted using Video Complexity Analyzer (VCA) [8]: ● average texture energy (EY), ● average gradient of the luma texture energy (h) ● average luma brightness (LY), ● average chroma texture energy of U and V channels (EU and EV) ● average chroma brightness of U and V channels (LU and LV) [8]. [8] V. V. Menon, C. Feldmann, K. Schoeffmann, M. Ghanbari, and C. Timmerer, “Green Video Complexity Analysis for Efficient Encoding in Adaptive Video Streaming,” in First International ACM Green Multimedia Systems Workshop (GMSys ’23), 2023.
- 9. Optimized resolution prediction 9 [9] V. V. Menon et al. Content-Adaptive Variable Framerate Encoding Scheme for Green Live Streaming. 2023. arXiv: 2311.08074 [cs.MM]. url: https://arxiv.org/pdf/2311.08074. [10] V. V. Menon et al. “Energy-Efficient Multi-Codec Bitrate Ladder Estimation for Adaptive Video Streaming”. In: 2023 International Conference on Visual Communications and Image Processing (VCIP). 2023.. Modeling • The perceptual quality and encoding time of the representation (rt , bt ) rely on the extracted video complexity features, encoding resolution, and target bitrate [9, 10]: • A higher resolution, and/or bitrate may improve the quality and increase the file size of the encoded video segment. • Similarly, a higher resolution, and/or bitrate can increase the encoding duration.
- 10. Optimized resolution prediction 10 [11] C. R. Helmrich et al., “Information on and analysis of the VVC encoders in the SDR UHD verification test,” in WG 05 MPEG Joint Video Coding Team(s) with ITU-T SG 16, document JVET-T0103, Oct. 2020. [12] M. Wien and V. Baroncini, “Report on VVC compression performance verification testing in the SDR UHD Random Access Category,” in WG 05 MPEG Joint Video Coding Team(s) with ITU-T SG 16, document JVET-T0097, Oct. 2020. [13] C. R. Helmrich et al., “A study of the extended perceptually weighted peak signal-to-noise ratio (XPSNR) for video compression with different resolutions and bit depths,” in ITU Journal: ICT Discoveries, vol. 3, May 2020. Optimization • LADRE optimizes the perceptual quality (in terms of XPSNR) of encoded video segments while adhering to real-time processing constraints. The optimization function is: We consider XPSNR as the perceptual quality measure instead of the popular VMAF. • The correlation of XPSNR with subjective quality scores is higher than VMAF for VVC-coded UHD bitstreams [11,12,13].
- 11. Optimized rate factor prediction 11 [14] V. V Menon, H. Amirpour, M. Ghanbari, and C. Timmerer, “ETPS: Efficient Two-Pass Encoding Scheme for Adaptive Live Streaming,” in Proc. IEEE Int. Conf. Image Process. (ICIP), Bordeaux, Oct. 2022. Modeling • Predicting the rate factor helps ensure consistent video quality throughout the stream. • It allows the encoder to allocate bits judiciously, preventing under-allocation (resulting in poor quality) or over-allocation (wasting bandwidth) of bits for encoding [14]. • The mathematical formulation of the rate factor optimization to yield a bitrate as close to the target bitrate as possible can be expressed as follows:. Optimization
- 12. Experimental design 12 Experimental parameters Table: Experimental parameters used to evaluate LADRE. • The Inter-4K dataset [15] is used to validate the performance. • The sequences are encoded using VVenC v1.10 [16] using preset 0 (faster). • The spatiotemporal features are extracted using VCA v2.0. • LADRE uses random forest regression models [17] to predict XPSNR, encoding time, and rate factor trained for each supported resolution. • Viewing resolution of UHD is assumed. [15] A. Stergiou and R. Poppe, “AdaPool: Exponential Adaptive Pooling for Information-Retaining Downsampling,” in IEEE Transactions on Image Processing, vol. 32, 2023, pp. 251–266. [16] A. Wieckowski, et. al., “VVenC: An Open And Optimized VVC Encoder Implementation,” in Proc. IEEE International Conference on Multimedia Expo Workshops (ICMEW), pp. 1–2. [17] L. Breiman, “Random Forests,” Machine Learning, vol. 45, 2001.
- 13. Experimental design 13 Benchmark schemes Table: An example fixed bitrate ladder, i.e., set of bitrate-resolution pairs. Source: [18]. 1. Default: This method employs a fixed set of bitrate- resolution pairs. We use the HLS bitrate ladder specified in the Apple authoring specifications [18] as the fixed set of bitrate-resolution pairs. 2. OPTE: This scheme predicts optimized resolution, which yields the highest XPSNR for a given target bitrate. [18] Apple Inc., “HLS Authoring Specification for Apple Devices.” [Online]. Available: https://developer.apple.com/documentation/http-live-streaming/hls-authoring-specification-for-apple- devices
- 14. Experimental results 14 Figure: RD curves and encoding times of representative video sequences (segments) using default encoding (blue line), OPTE (purple line), LADRE (red line).
- 15. Experimental results 15 Table: Average results of the encoding schemes compared to the Default encoding. ● Coding efficiency (in terms of Bjøntegaard Delta [19] rates), and encoding times decrease as the encoding time constraint is decreased. ● OPTE yields the highest coding efficiency (maximizes XPSNR with no encoding time constraint). ● We yield storage reduction using LADRE. ● Some high bitrate representations are eliminated because it cannot be encoded with the target maximum encoding time. ● The maximum encoding time for a video segment in parallel encoding environment is shown in the table below (last column). It is below the desired threshold. [19] HSTP-VID-WPOM, “Working practices using objective metrics for evaluation of video coding efficiency experiments,” International Telecommunication Union, 2020. [Online]. Available: http://handle.itu.int/11.1002/pub/8160e8da-en
- 16. Conclusions 16 ● We proposed an encoding latency-aware dynamic resolution per-title encoding scheme (LADRE) for adaptive streaming applications. ● LADRE includes an optimized resolution prediction, which uses random forest-based models to estimate bitrate-resolution-rate factor triples for a given video segment based on its spatial and temporal characteristics. ● On average, LADRE, with a target encoding time constraint of 200 s, yields bitrate savings of 10.25% and 12.03% to maintain the same PSNR and XPSNR, respectively, compared to the reference HLS bitrate ladder with a negligible additional latency in streaming. ● Furthermore, an average decrease of 84.17% in encoding energy consumption is observed.
- 17. Reproducibility 17 LADRE is implemented in a Python-based framework: ● Available: https://github.com/PhoenixVideo/QADRA ● Monotonicity of XPSNR and QP predictions ● Addressed in QADRA using a cascading approach instead of using a single regression model.
- 18. 18