Inventory Demand Forecasting using Machine Learning in R

The dataset used is from Mexican multinational company, Grupo Bimbo. The delivery chain is present in countries of America, Europe, Asia, and Africa. It has an annual sales volume of 15 billion dollars. Grupo Bimbo delivers fresh bakery products to 1 million stores along its 45,000 routes across Mexico. There are five datasets available.

 

train.csv: It contains data available for training. The dataset contains about 7.4 billion entries over 11 features.

train.csv: It contains data available for testing.

cliente_tabla.csv: It contains the data of 93 million clients over 2 different features.

producto_tabla.csv : It contains the data for 2592 products.

town_state.csv : It contains the data for 790 towns 

Data set : https://www.kaggle.com/c/grupo-bimbo-inventory-demand/data

Language : R

Libraries : dplyr, ggplot2, caTools, xgboost, data.table, Matrix, lightgbm, gbm, caret, zoo, Data Combine, e1071

 

Exploratory data analysis (EDA)

Data visualization

Inference about features

Feature engineering

 

2. Data cleaning (outlier/missing values/categorical)

Missing value detection

Replacing special characters in data

 

3. Feature Engineering

Adding lag columns for target variable

Adding moving average columns for target variable

 

4. Model building on training data

XGBoost

GBM

SVM

 

5. Model validation

RMSE

 

6. Conclusion

Understanding the Business context and objective

Data Cleaning

Inference about data

Feature Engineering

Importing the dataset and importing libraries

Train and test split for model validation

Significance of Train Test split

Different Evaluation parameters and their meaning

Understanding Bagging and Boosting

Understanding XGBoost

Hyperparameter Tuning

Understanding GBM and light GBM

Understanding SVM

Building XGBoost model

Building GBM

Building SVM model