Modeling and Simulation of Calibrated Quantile Forecasters
Implementation plan:
*************************
Step 1: Initially, we Collect and load data from 4G LTE Speed trained Dataset .
Step 2: Then we Preprocess the data by cleaning missing values, adding basic features, and preparing time-series windows.
Step 3: Next, we Build forecasting models using TCN and Transformer networks for predicting quantiles.
Step 4: Next, we implement Federated Learning (fedAvg or Fedprox) using the Flower framework and train models on multiple client nodes.
Step 5: Next, we Apply Conformal Quantile Regression (CQR) method to calibrate and improve prediction intervals .
Step 6: Finally, we plot performance metrics for the following
6.1: Number of epochs vs. Accuracy (%)
6.2: Number of epochs vs. Loss (%)
6.3: Number of epochs vs.Pinball Loss (Mbps)
6.4: Number of traces vs. PICP coverage (%)
6.5: Number of traces vs. Mean Interval Width (Mbps)
Software Requirements:
***************************
1. Development Tool: Python 3.11.4 or above
2. Operating System: Windows 10 (64-bit) or above
Dataset:
*********
Link: https://www.kaggle.com/datasets/aeryss/lte-dataset
Note:
******
1) If the proposed plan does not fully align with your requirements, please provide all necessary details—including steps, parameters, models, and expected outcomes—in advance. Kindly ensure that any missing configurations or specifications are clearly outlined in the plan before confirming.
2) If there’s no built-in solution for what the project needs, we can always turn to reference models, customize our own, different math models or write the code ourselves to fulfil the process.
3) If the plan satisfies your requirement, Please confirm with us.
4) Project based on Simulation only.
5) If you have any dataset to change,kindly provide us before implementing it.
