UCB FinTech Bootcamp Module 11 Challenge

Mercado Libre Traffic Analysis

Introduction

The Bootcamp Module 11 Challenge - "Mercado Libre" Traffic Analysis requires the preparation of a strategy to help the company grow by using search traffic predictions as an indicator for stock trading opportunities and sales through the site.

Goals and Objectives

Part I. : Analyze the hourly Google Search traffic data to find patterns.

Part II. : Mine the Search traffic data to determine time of the day (hourly) and day of the week (daily) seasonality.

Part III. : Relate the Search traffic data to Mercado Libre stock price patterns over time.

Part IV. : Create a Search traffic forecast/time series using Prophet.

Part V. : Create a site Sales/Revenue forecast using Prophet.

Technologies and Tools

The following list includes the main technologies and tools using during the preparation and deployment of the solution:

1. Python - Programming language used to code the solution. Version 3.7.13 was used. Required libraries and frames listed in the Installation section below

2. GitHub - Reposotory for code deployment, version management and documentation of the presented solution

3. Jupyter Labs - IDE tool for coding, code testing/debugging and solution documentation. Version V3.4.4 was used

4. Git Bash console - Local console used to test the coded solution and sync wiht GitHub Version 2.40.0.windows.1 was utilized

5. Slack - Collaboration tool to communicate and brainstorm with other FinTech Bootcamp participants

6. Operative System - This solution was prepared in a PC running Windows 11 v H22

For additional details, please refer to the watermark output at the bottom of the Jupyter Notebook

Installation Guide

1. prophet : Prophet is a procedure for forecasting time series data based on an additive model where non-linear trends are fit with yearly, weekly, and daily seasonality, plus holiday effects. It works best with time series that have strong seasonal effects and several seasons of historical data. Prophet is robust to missing data and shifts in the trend, and typically handles outliers well. Prophet is open source software released by Facebook’s Core Data Science team. It is available for download on CRAN and PyPI. Install in your environment following the steps below:

1.1. Open the GitBash terminal

1.2 Type the following command and press Enter:

pip install prophet

Note: The above installation instructions are for the Prophet package to work locally in a Windows computer; if Prophet does not work locally, virtual environment options, such as Google Colab can be utilized.

2. hvplot : hvplot is a Python library that provides a high-level interface for creating interactive and visually appealing visualizations. It is built on top of HoloViews and provides a simplified API for generating plots directly from Pandas DataFrames, GeoPandas DataFrames, xarray datasets, and other data structures commonly used in the PyData ecosystem. To install, follow these steps:

2.1. Open the GitBash terminal

2.2 Type the following command and press Enter:

pip install hvplot

Solution Structure

The 11.Mercado_Libre repository in GitHub contains the solution components. The repository consists of the following folders and contents as described below:

1. Resources : Contains the .csv files with the Mercado Libre related stats for Google searches, Stock close prices and site Sales, which serve as the basis for the analysis.

2. gitignore : Instructions for which files/file types to exclude from the sync process between GitHub and the local environment.

3. README.md : The present file containing this outline of the challenge and its components.

4. mercado_libre_forecasting_prophet.ipynb : This is the Jupyter Notebook with the code for the core challenge solution, analysis and conclusions based on the data available in the Resources folder.

User Instructions

1. Launch the Jupyter Lab from the GitBash terminal using the following command line:

jupyter lab

2. From the Jupyter Lab terminal, navigate to the location of the mercado_libre_forecasting_prophet.ipynb notebook and open it

3. Reset the Kernel and

4. Run the steps from the top and in sequence verifying the results as per the comments in the notebook, including outputs on the screen such as dataframe listing/on screen printing and generated graphs/plots (e.g. heathmap plots for Search Trends by time period, etc.) as well as answers to the specific questions in the challenge.

Credits

Prepared by Luis Torres

LUTOV001

June 2023

LUTOV001/11.Mercado_Libre

UCB FinTech Bootcamp Module 11 Challenge

Mercado Libre Traffic Analysis

Introduction

The Bootcamp Module 11 Challenge - "Mercado Libre" Traffic Analysis requires the preparation of a strategy to help the company grow by using search traffic predictions as an indicator for stock trading opportunities and sales through the site.

Goals and Objectives

Part I. : Analyze the hourly Google Search traffic data to find patterns.

Part II. : Mine the Search traffic data to determine time of the day (hourly) and day of the week (daily) seasonality.

Part III. : Relate the Search traffic data to Mercado Libre stock price patterns over time.

Part IV. : Create a Search traffic forecast/time series using Prophet.

Part V. : Create a site Sales/Revenue forecast using Prophet.

Technologies and Tools

The following list includes the main technologies and tools using during the preparation and deployment of the solution:

1. Python - Programming language used to code the solution. Version 3.7.13 was used. Required libraries and frames listed in the Installation section below

2. GitHub - Reposotory for code deployment, version management and documentation of the presented solution

3. Jupyter Labs - IDE tool for coding, code testing/debugging and solution documentation. Version V3.4.4 was used

4. Git Bash console - Local console used to test the coded solution and sync wiht GitHub Version 2.40.0.windows.1 was utilized

5. Slack - Collaboration tool to communicate and brainstorm with other FinTech Bootcamp participants

6. Operative System - This solution was prepared in a PC running Windows 11 v H22

For additional details, please refer to the watermark output at the bottom of the Jupyter Notebook

Installation Guide

1.1. Open the GitBash terminal

1.2 Type the following command and press Enter:

Note: The above installation instructions are for the Prophet package to work locally in a Windows computer; if Prophet does not work locally, virtual environment options, such as Google Colab can be utilized.

2.1. Open the GitBash terminal

2.2 Type the following command and press Enter:

Solution Structure

The 11.Mercado_Libre repository in GitHub contains the solution components. The repository consists of the following folders and contents as described below:

1. Resources : Contains the .csv files with the Mercado Libre related stats for Google searches, Stock close prices and site Sales, which serve as the basis for the analysis.

2. gitignore : Instructions for which files/file types to exclude from the sync process between GitHub and the local environment.

3. README.md : The present file containing this outline of the challenge and its components.

4. mercado_libre_forecasting_prophet.ipynb : This is the Jupyter Notebook with the code for the core challenge solution, analysis and conclusions based on the data available in the Resources folder.

User Instructions

1. Launch the Jupyter Lab from the GitBash terminal using the following command line:

2. From the Jupyter Lab terminal, navigate to the location of the mercado_libre_forecasting_prophet.ipynb notebook and open it

3. Reset the Kernel and

Credits

Prepared by Luis Torres

LUTOV001

June 2023