Important : Due to cache usage limitations on free hosting platforms , this application is hosted locally & not on web servers . I have pasted the screenshot showing how this application works. Detailed code is available in repository .
Shortages – Is one such word which every stakeholder in the supply chain industry comes across – be it a Buyer / Logistics Coordinator / Production Supervisor.
Shortages are inevitable , however those can certainly be reduced with the help of machine learning based forecasting models.
Forecasting is an art of being less wrong , In this next use case of #SCMunfolds , I would like to showcase a prototype model for Demand Forecasting using Machine Learning.
This model compares various regression / time series techniques to predict the demands based on the historical data patterns predicting the best model to be used with highest level of accuracy amongst all.
Being a prototype , a very limited data is being displayed in this model , we can however explore it to the level desired.
You can upload a simple CSV file having date in first column & respective value in next consecutive column.
Library Used : PyCaret , PyCaret is a powerful library which has applications across TimeSeries , Classification & Regression
Input Screen : Select the slider to decide how many days of data you want to predict.
No. of Days I have selected here are 30 , Click on 'Give a try with our example dataset' , model will start predicting the most suitable model.
Output matrix shows the comparison of performance of all the models : Models having lowest error values are highlighted
Output Screen : Based on the output of compariosn matrix : ARIMA model is the most suitable model for the example dataset . Plotting the values predicted by model on 'Test Data' . Predicted values are shown in Orange color
Diagnostics : This is how ARIMA model is performing. We can infer many a insights from this data - which we can further use for analysing the business.
Predictions : As ARIMA has best accuracy , using this model to predict the future values . Future values are shown in Blue Color
