This is a project to takle a challege proposed by the Magazine Luiza team. Predict the demand of products for next N months based on a history of previous sales.
If you want more information you can go to the Experiments Journal.
First you download the data needed and unpack it.
$ bash -i download_data.sh
Then you build the environment:
$ bash -i build_env.sh
├── build_env.sh
├── download_data.sh
├── README.md
├── data # Datasets and intermediate data files.
│ |
│ ├── clean_step1.csv # During the EDAs, it was the first clean step.
│ ├── clean_step2.csv # During the EDAs, it was the last clean step.
│ |
│ ├── desafio.csv # Data provided by the company.
│ |
│ ├── forecast_dataset_exp11.csv # Dataset used for prediction experiments.
│ ├── forecast_dataset_exp12.csv # Dataset used for prediction experiments.
│ ├── forecast_dataset_exp13.csv # Dataset used for prediction experiments.
│ ├── forecast_dataset_exp14.csv # Dataset used for prediction experiments.
│ ├── forecast_dataset_exp1.csv # Dataset used for prediction experiments.
│ |
│ ├── output_desafio_exp12.csv # Dataset with prediction predictions.
│ ├── output_desafio_exp13.csv # Dataset with prediction predictions.
│ ├── output_desafio_exp14.csv # Dataset with prediction predictions.
│ |
│ ├── products_features_exp2.csv # Dataset with product embeddings.
│ ├── products_features_exp3.csv # Dataset with product embeddings.
│ ├── products_features_exp4.csv # Dataset with product embeddings.
│ ├── products_features_exp5.csv # Dataset with product embeddings.
│ |
│ └── train_val_test.csv # Train test split by year_month
|
├── docs # Documentation and notes
│ ├── Desafio Ciência de Dados - Previsão de Demanda.pdf
│ └── notes.md
│
├── notebooks # Experiments - Check the README.md
│ ├── README.md # Experiment Journal
│ ├── 1. EDA Orders.ipynb
│ ├── 1.1 EDA Orders.ipynb
│ ├── 2. EDA Products.ipynb
│ ├── 3. EDA Correlations.ipynb
│ ├── 4. Clustering - Feature Engineering.ipynb
│ ├── 6. Prediction - Exp 1 DatasetPrep.ipynb
│ ├── 5. Clustering - Exp 1.ipynb
│ ├── 5. Clustering - Exp 2.ipynb
│ ├── 5. Clustering - Exp 3.ipynb
│ ├── 5. Clustering - Exp 4.ipynb
│ ├── 5. Clustering - Exp 5.ipynb
│ ├── 6. Prediction - Exp 1.ipynb
│ ├── 6. Prediction - Exp 2.ipynb
│ ├── 6. Prediction - Exp 4.ipynb
│ ├── 6. Prediction - Exp 5.ipynb
│ ├── 6. Prediction - Exp 6.ipynb
│ ├── 6. Prediction - Exp 7.ipynb
│ ├── 6. Prediction - Exp 8.ipynb
│ ├── 6. Prediction - Exp 9.ipynb
│ ├── 7. Prediction - Exp 10 - Quantile Regression.ipynb
│ ├── 7. Prediction - Exp 11 DatasetPrep.ipynb
│ ├── 7. Prediction - Exp 11.ipynb
│ ├── 7. Prediction - Exp 12 DatasetPrep.ipynb
│ ├── 7. Prediction - Exp 12 - Finetunning.ipynb
│ ├── 7. Prediction - Exp 12.ipynb
│ ├── 7. Prediction - Exp 13 DatasetPrep.ipynb
│ ├── 7. Prediction - Exp 13 - Finetunning.ipynb
│ ├── 7. Prediction - Exp 13.ipynb
│ ├── 7. Prediction - Exp 14 DatasetPrep.ipynb
│ ├── 7. Prediction - Exp 14 - Finetunning.ipynb
│ └── 7. Prediction - Exp 14.ipynb
|
└── src # Helper code
├── metrics.py # Metrics and some evaluation for re-use.
├── mlp_quantile.py # Neural network with the titled_loss (quantiles)
├── optimizer.py # Optimizer used for Clustering - Elbow Method
└── visualization.py # Visualization for regressions
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent