How vihaanbishnoi/automl works
This is a Streamlit time-series regression recommender: users upload a CSV, choose its time/entity columns and prediction target/features, then run preprocessing, windowing and model training. The pipeline evaluates classical, deep-learning and (when eligible) ARIMA forecasts, sorts results by RMSE, and the UI displays metrics and the top recommendation. Edges reflect shown calls or SOURCE INDEX references; unsampled module behavior is not inferred.
Main entry points: model_recommender.py Recommendation pipeline · deep_learning.py Deep learning · app.py Streamlit app · classical.py Classical training
Components 15
User interface
- Streamlit app
app.py
Time-series preparation
- Data cleaning
data_utils.py - Window creation
model_utils.py
Model training
- Recommendation pipeline
services/model_recommender.py - Classical training
trainers/classical.py - Deep learning
trainers/deep_learning.py - Statistical training
trainers/statistical.py - Random forest
models/random_forest.py - LSTM model
models/lstm.py - GRU model
models/gru.py - 1D CNN model
models/cnn1d.py - ARIMA model
models/arima.py
Evaluation and results
- Regression metrics
evaluation_utils.py - Results display
visualization/plots.py
Other components
- CSV user
Connections 16
- CSV user → Streamlit app: uploads CSV
app.py - Streamlit app → Recommendation pipeline: runs pipeline
app.py - Recommendation pipeline → Data cleaning: cleans data
services/model_recommender.py - Recommendation pipeline → Window creation: creates windows
services/model_recommender.py - Recommendation pipeline → Classical training: trains models
services/model_recommender.py - Recommendation pipeline → Deep learning: trains models
services/model_recommender.py - Recommendation pipeline → Statistical training: trains models
services/model_recommender.py - Classical training → Random forest: trains model
trainers/classical.py - Classical training → Regression metrics: evaluates predictions
trainers/classical.py - Deep learning → LSTM model: builds model
trainers/deep_learning.py - Deep learning → GRU model: builds model
trainers/deep_learning.py - Deep learning → 1D CNN model: builds model
trainers/deep_learning.py - Deep learning → Regression metrics: evaluates predictions
trainers/deep_learning.py - Statistical training → ARIMA model: fits and forecasts
trainers/statistical.py - Statistical training → Regression metrics: evaluates forecasts
trainers/statistical.py - Streamlit app → Results display: displays results
app.py