· Overview:

• Downloaded 5 datasets from Kaggle(Footlocker, ASOS, H&M, Mango and Zara). -

• Webscrapped "Massimo Dutti" website and downloaded the relevant data for both men and women.

• Imported all the necessary libraries.

• Created a dataframe for each dataset.

• Cleaned and organize each dataset separately.

• Calculated the mean price for each dataset.

• Created csv for each dataset after being cleaned (for later usage in Tableau).

• Concatenated the 6 datasets (brands_df) and drop the necessary columns.

• Combined data from various columns into new consolidated columns, then clean up redundant columns to refine the Dataframe.

• Renamed brands_df as df_cleaned and drop innecesary columns.

• Managed all the null values in different ways (dropping columns, filling with mode, filling with strings).

• Created a csv for df_cleaned.

  · EDA:

• Computed mean price for each group per section and display it in a bar chart.

• Computed average price for each brand and display it in a bar chart.

• Displayed in a barchart and dataframe for Footlocker's top 10 brands.

• Created a pie chart representing the relationship between sales volume and promotion in Zara.

• Displayed in a barchart H&M's top 10 colors.

• Displayed in a barchart ASOS top 10 colors.

• Displayed in a barchart and dataframe for Mango's top 10 categories by product.

• Displayed in a barchart and dataframe for Massimo Duti's top 10 categories by average price.

• Created a histogram for the price of the whole dataset (df_cleaned).

  · HYPOTHESIS TESTING:

• Computed One Sample T-Test, with its corresponding hypotheses, significance level and its values well defined and calculated in order to compare it with alpha and reject or accept H0.

• Computed two more tests but this time Two Sample T-Test with independent variables.

• The last test was a Proportion Z-Test with both hypotheses and values well defined in order to compare it and reject H0.

  · MACHINE LEARNING: (Different notebook)

• Imported the necessary libraries.

• Imported and defined as "brands" brands_csv.

• Created a Correlation Matrix dropping all the columns and just taking into account 'price', 'promotion' and 'section' .

• Applyed feature scaling to numerical columns (price).

• Train-test split process, defining the feature and target.

• Linear regression model and its corresponding evaluation (R², MAE and RMSE).

• Decision Tree model and its corresponding evaluation (R², MAE and RMSE).

• Random Forest model and its corresponding evaluation (R², MAE and RMSE).

• Gradient Boosting and its corresponding evaluation (R², MAE and RMSE).

• Hyperparameter Tuning for Random Forest to look for the best params and score.

• Hyperparameter Tuning for Gradient Boosting to look for the best params and score.

• Compared the two hyperparameters tunning and create a csv with the best model (Gradient Boosting).

• Displayed the results of the actual price and it's prediction.

  · SQL:

• Imported both "brands.csv" and "gradient_boosting_predictions.csv" into SQL workbench.

• Created a database for each csv.

• Confirmed that each csv was well imported and display their corresponding columns and values.

• Saved both SQL files.

  · TABLEAU:

• Imported the 6 cleaned csv into different Data Sources.

• Analyzed each database in two different ways. (we have changed from dimension to measure some of the fields when needed).

• Created a dashboard for each database showing both sheets with their corresponding filters.

  · APP CREATION:

• Created a copy of the "brands_final_project.ipynb" file and converted it into a Python script for improved modularity and usability.

• Defined a "load_data()" function within this Python file to handle loading the entire "brands.csv" dataset.

• Developed an additional Python file dedicated to app creation using Streamlit.

• Imported Streamlit to build an interactive user interface.

• Imported the data-handling script to utilize the "load_data()" function.

• Implemented filter options to enable users to filter brands by key attributes: Brand, Section, Price Range, and Category.

• Executed the Streamlit script in the terminal, generating a link to our fully functional app, complete with filtering capabilities for user exploration of the dataset. This README section provides a clear summary of the project structure, functionality, and app deployment steps.