Skip to content

Time series prediction system in R (RStudio) for given real-time e-commerce dataset of thousands of products, customers, and categories with the help of data mining algorithms (ARIMA, Holt Winter, STL, ETS).

Notifications You must be signed in to change notification settings

CG-R8/Sales-time-series-prediction

Repository files navigation

Time-Series-Forecasting

Description:

  • A sales data of 118 days has been provided for 100 key product ids. The goal of this project is to predict next 28 days sales of 100 key product ids. After studying the data, it was found that the data had no seasonality. So I came to the conclusion of using Seasonal Decomposition of Time Series by Loess model and Error Trend Seasonal (ETS) method.
  • This repository contains data files provided, run.R and data_trend_pred.R
  • run.R has step by step function calls for execution
  • data_trend_pred.R contains the functions performing reading and writing of data and forecast function.

Requirement:

Installation:

  • Install the R studio from the link provided.
  • Install the forecast library from package installer found in R studio.

Usage:

  • Open the R studio change the working directory where all the R files and data files are present.
  • Run the run.R in R studio by including the file using source("run.R") command.
  • This will run the run.R file by step by step executing all the commands present in the file.
  • Finally it will write the output in output.txt file and store in the working directory.

About

Time series prediction system in R (RStudio) for given real-time e-commerce dataset of thousands of products, customers, and categories with the help of data mining algorithms (ARIMA, Holt Winter, STL, ETS).

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages