Scribd
Upload
38 views11 pages

Optimizing Sales Forecasting - A Comprehensive Analysis

The document outlines a project at ABC Inc aimed at improving sales forecasting accuracy through data preprocessing, exploratory data analysis, and model building. It details the methodologies used, including various regression techniques and hyperparameter tuning, ultimately selecting the Random Forest Regressor as the best model based on evaluation metrics. The conclusion emphasizes the importance of deploying the model for real-time forecasting and ongoing monitoring.

Uploaded by

anusha
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
38 views11 pages

Optimizing Sales Forecasting - A Comprehensive Analysis

The document outlines a project at ABC Inc aimed at improving sales forecasting accuracy through data preprocessing, exploratory data analysis, and model building. It details the methodologies used, including various regression techniques and hyperparameter tuning, ultimately selecting the Random Forest Regressor as the best model based on evaluation metrics. The conclusion emphasizes the importance of deploying the model for real-time forecasting and ongoing monitoring.

Uploaded by

anusha
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 11

Sales Forecasting and Analysis

Improving Sales Forecasting Accuracy at ABC Inc


Agenda
Introduction
Data Preprocessing
Exploratory Data Analysis (EDA)
Model Building
Model Evaluation
Conclusion and Next Steps
Introduction

Company Overview: ABC Inc, a US-based specialty retailer of crafts and fabrics.

Problem Statement: Improving sales forecasting accuracy for better budget


planning.

Importance: Accurate forecasting informs budget decisions and inventory


management.
Link to the python code
Data Preprocessing
Data Sources: Sales Data, Promotion Calendar, Product Master.
Steps:
Outlier Detection and Treatment.
Missing Value Treatment.
Combining and Cleaning Datasets.
Link to the python code
Exploratory Data Analysis (EDA)
Visualizations:

Sales Patterns Over Time.

Seasonality and Trends.

Key Insights:

Seasonal Decomposition.

Offer Effects.

link to the python code


Seasonality and Trend Analysis
Model Building
Model Selection:
RandomForestRegressor, GradientBoostingRegressor.
Feature Engineering:
Product, Sub_Category, Category, Unit_Retail_Price, Offer, Offer_Type, Effective_Price,
Year, Month, Day, DayOfWeek, Quarter, WeekOfYear, IsWeekend, DayOfYear,
WeekdayName, SeasonNameCode
Creating relevant features.
Date Split: get Year, Month, Day, DayOfWeek, Quarter, WeekOfYear, IsWeekend,
DayOfYear, WeekdayName, SeasonNameCode encoding them in numeric form
Training and Validation sets: splitting 80 to 20 ratio
Hyperparameter Tuning: GridSearchCV for best model parameters.
link to the python code
Model Building (Revised)
Expanding the Regression Toolkit for Enhanced Performance.
Introduction to Additional Regression Techniques:
Lasso Regression ,Ridge Regression, Elastic Net Regression, Support Vector Regression
(SVR), Principal Component Analysis (PCA) ,Kernel Ridge Regression, Gaussian Process
Regressor ,MLP Regressor
Diverse Set of Techniques:
Aiming to further reduce errors and improve forecasting accuracy.
Rigorous Hyperparameter Tuning:
Employing GridSearchCV for each technique to identify optimal parameters.
Link to the python code
Model Evaluation
Evaluation Metrics: Random Forest Regressor
Mean Absolute Error (MAE): 19.488
Model Comparison:between Random Forest Regressor and Gradient
boosted tree
Performance on validation data.
Best Model Selection:Random Forest Regressor
Chosen model based on lowest MAE: Random Forest Regressor

Link to the python code


Conclusion
Summary of Findings:

Improved Sales Forecasting Accuracy.

Next Steps:

Deploying the model for real-time forecasting.

Continual model monitoring and updates.


Link to the python code
THANK YOU

You might also like