Dataset provided with historical sales data for 45 stores located in different region search store contains a number of departments. The company also runs several promotional markdown events throughout the year. These markdowns precede prominent holidays, the four largest of which are the Super Bowl, Labor Day, Thanksgiving, and Christmas. The weeks including these holidays are weighted five times higher in the evaluation than non-holiday weeks.
One challenge of modeling retail data is the need to make decisions based on limited history. Holidays and select major events come once a year, and so does the chance to see how strategic decisions impacted the bottom line. In addition, markdowns are known to affect sales the challenge is to predict which departments will be affected and to what extent.
- Over here we have 3 CSV file, named as Store_Details, Business_Data and Sales_History.
- Anonymized information about 45 stores indicating store type, Address, Location and size of store.
Contains additional data related to the store, department, and regional activity for the given dates.
- Store - the store number
- Date - the week
- Temperature - Average temperature of surrounding region of store location.
- Fuel_Price –Average price of gas per gallon.
- MarkDown1-5 - Anonymized data related to promotional markdowns. MarkDown data is not available for all stores all the time. Any missing value is marked with an “NA”.
- CPI - The Consumer Price Index (CPI) is a measure of the average change in prices over time that consumers pay for a weighted averaged basket of goods and services.
- Unemployment_Rate - the unemployment rate of the region where the store is located.
- Holiday - whether the week is a special holiday week
Historical sales data of each store from 30th April 2017 to 25th October 2019. Within this tab you will find the following fields:
- Store - the store number
- Department - the department number
- Date - the week
- Total_Sales –Total sales for the given department in the given store
- Holiday - whether the week is a special holiday wee
- pridected total sales dataset are the predicted sales data.
- Understand the data variables properly. Checking the variable description to understand the data properly.
- Clean the data: Cleaning the data, that is, filling the missing values (if any), treat the outliers (or odd values), etc. Ensuring each variable’s data is as per the nature of the variable (e.g. Date field should contain only date values – can extract year, month and day of the week, and numeric column should be formatted as numeric, etc.).
- Conduct EDA (Exploratory Data Analysis) on the cleaned Data: Summarized and explore the data and then decide my strategy.
- Uni-variate and Bi-variate Analysis: Checking the distribution of independent variables and
also compare them with the dependent variable. - Feature Engineering: Creating new meaningful features based on the existing features by applying some aggregation functions on them.
- Hypothesis Testing: Hypothesis testing in statistics is a way for you to test the results of a survey or experiment to see if you have meaningful results. Given a brief summary of the data and a summary of the results of your statistical test.
- Identify the most important variables (or data parameters) that affect the final decision: Identify the impact of each variable on the final result graphically (correlation / scatter plots, regression plots, etc.). Keep those variables that affect the final outcome.
- Develop and Validate Samples: Divide samples into 2 parts: Development Sample (80%) & Validation Sample (20%). Building analysis model using the Development Sample, and validate it on the validation sample and then predict on test sample.
- Model Building: Analyze the dependent variable and decide which technique out of regression or classification to use and hence build the model.
- Improving model accuracy: We know that machine learning algorithms are driven by parameters. These parameters majorly influence the outcome of learning process. So, find the optimum value for each parameter to improve the accuracy of the model and repeat this process with a number of well performing models.
- Model Comparison: Comparing the each model with other similar models and then choose that model which give highest accuracy. But it is not necessary that higher accuracy models always perform better (for unseen data points). So, find the right accuracy of the model, you must use cross validation technique before finalizing the model.
- numpy
- matplotlib
- scikit-learn
- seaborn
- pyhthon 3.9
Just run jupyter notebook in terminal and it will run in your browser.
Install Jupyter here i've you haven't.
- Dataset are already uploded in this repo https://github.com/rahkum96/TARGET-STORE-SALES-PREDICTION