layout

By default, the <title> tag for this page is determined by the title field (in the Content group). You can override the tag value here.

metaTitle

media

bottomSections

By default, the description tag for this post is taken from the Excerpt field. You can override the tag value here.

metaDescription

By default, the "og:image" meta tag set for social sharing this post points to the Featured Image field (in the Content group). You can override that image here.

socialImage

markdown_content

If enabled, the title suffix defined in the site configuration is appended to the title tag of this page.

addTitleSuffix

To add or override any meta tag for this page, add entries to this list. Entries defined here take precedence over any other defaults.

metaTags

title

colors

date

author

excerpt

featuredImage

Scraping & EDA Amsterdam Restaurants and Cafés on Google Maps

In this project, I tried to create a model that can predict the price of a laptop based on the criteria of the desired laptop by using the data of the available laptops that I collected from the Digikala website, the largest e-commerce site in Iran.

[Project Github Repository](https://github.com/meysamraz/laptop-price-prediction-end-to-end-project-using-ecommerce-website-data)

[Project Demo](https://loptop-price-prediction.herokuapp.com/)

***Note: Since the price of laptops in Iran is constantly changing, this model is trained based on the data of laptops available on 2022-11-06 (1401-8-15) in Digikala. And in the coming years, it may not have an accurate prediction, and the solution to this problem is to update the model with the most up-to-date data.***

# Project Overview :

## 1 - Collect Data

To collect my data, I used Digikala's secret api. I was able to collect the data I wanted (available laptops with prices) with a simple fitler.

*   #### Collect Laptop main data
    *   id
    *   title\_fa
    *   title\_en
    *   price
    *   image\_url
    *   brand

*   #### Collect Laptop details data
    *   cpu manufacturer
    *   cpu series
    *   cpu model
    *   ram
    *   ram type
    *   internal storage
    *   internal storage type
    *   gpu manufacturer
    *   gpu model
    *   screen resolution
    *   ports

*   #### Merge Collected data

*   #### Remove duplicated rows

*   #### Save data into csv file

## 2 - Take a look at data :

After collecting the data, I started checking the collected data to make sure it was collected correctly

*   #### Check shape of data
*   #### Check is there any null value
*   #### Check data types
*   #### Check number of unique values in each column

## 3 - Cleaning data

Like all machine learning projects, the data doesn't arrive perfect and ready for prediction. At this point, I started cleaning the collected data.

*   #### Convert brands name from persian to english
*   #### Convert ram from persian to english digits
*   #### Clean and convert internal storage to english
*   #### Convert and clean internal storage to english
*   #### Convert and clean laptops screen size
*   #### Clean laptops resolution

## 4 - EDA

![](/images/plot1.png)

For the next step, which is **Feature engineering**, it was necessary to get information about the data. In this step, I analyzed and explored the data.

*   #### Laptops price distribution
*   #### Number of laptops of each brand
*   #### Number of cpu of each cpu manufacturer
*   #### Number of laptops for each ram group
*   #### Number of laptops for each ram type group
*   #### Number of laptops with diffrent internal storage
*   #### Number of laptops for each internal storage group
*   #### Number of laptops with diffrent screen sizes
*   #### Number of laptop with diffrent screen resolution

## 5 - Feature Engineering

In this step, I prepared the features for training the model

*   #### Remove outliers base on laptops price using z score
*   #### Convert screen resolution to number
*   #### Extract Gaming brands from title (asus rog , acer nitro ...)
*   #### Remove brand with only one laptop
*   #### Extract clean gpu model from gpu model column
*   #### Remove laptops with only less than 3 model gpu
*   #### Label endcoding cleand gpu models
*   #### Convert internal storage from tb and gb to mg
*   #### Label encoding internal storage type
*   #### Convert ram from str to int
*   #### Extract port count
*   #### Label encoding ram type
*   #### Label encoding cpu series
*   #### One hot encoding brand - cpu manufacturer - gpu manufacturer (nominal categorical variables)

## 6 - Feature Selection

In this step, I chose the features needed to train the model

*   #### Check correlation
*   #### Mutual information regression

## 7 - Model training

![](/images/plot4.png)

## 8 - Hyperparameter tuning

In this step, I hyperparameter tuned the model that had the best r2 score

## 9 - Cross validation

![](/images/plot5.png)

## 10 - Save model

I pickeld model for use in the gui environment

## 11 - Website

To create website, I used streamlit formwork, a powerful formwork that allows me to create the desired user interface completely using Python.

## 12 - Deploy

I used [Heroku](https://www.heroku.com/) a cloud platform as a service which provide a free hosting to deploy my app on it. it's and amzaing platform gave me so much flexbilte to deploy your apps

## Libraries and FrameWorks used in the project

*   [streamlit](https://streamlit.io/)
*   [pandas](https://pandas.pydata.org/)
*   [numpy](https://numpy.org/)
*   [matplotlib](https://matplotlib.org/)
*   [seaborn](https://seaborn.pydata.org/)
*   [termcolor](https://pypi.org/project/termcolor/)
*   [scikit-learn](https://scikit-learn.org/)
*   [scipy](https://scipy.org/)
*   [xgboost](https://xgboost.readthedocs.io/)



Predicting Laptop Prices: An End-to-End Project Using E-commerce Website Data

[Project Github repository
](https://github.com/meysamraz/Snapp-ds-dashboard)

Snapp! is an Iranian service provider, allowing users to book a car and driver to transport them in a way similar to a taxi. Snapp is currently the most popular online taxi rental service in Iran, so I decided to analyze my use of this popular service.

### Project Overview :





#### 1 - Collecting Data :

Since Snapp did not have an api or source to receive data related to my travel history. I used a bot to collect my data from the snapp web app

#### What data did I collect ?

*   Number of travels
*   Total Hours of travels
*   Total kilometers traveled
*   Travel day title
*   Car that i traveld with
*   Travel cost
*   Travel date
*   trave time (hour)
*   Initial point of travel
*   Travel destination
*   How many star were my travel

#### 2 - Preprocess and Feature engineering collected data :

*   Drop canceld travels
*   Extract destination location from data
*   Extract origin city (translated in english)
*   Extract date and convert it form jalali (persain date unit into gregorian)
*   Translate car model in English
*   Convert travel cost from rial to toman and translate it (English)
*   Split Curreny from price
*   Drop uneeded columns
*   Save data into csv file

#### 3 - Creating the dashboard using Power BI

For creating dashboard i used Power BI an powerful tool and interactive data visualization software product developed by Microsoft

#### What i persented in my dashboard :

*   Total Expense of travels
*   Total Hours of travels
*   Total kilometers traveled
*   Total stars givin
*   Total cost of travels (dynamic in year and month)
*   Quality of my travels (How many stars were my tavels mostly ?)
*   Where was the most expensive travels for me ?
*   In which cities did I use snapp the most ?
*   My travels table (all of my travels (dynamic by sorting) )

#### Libraries and software used :

*   [Power BI](https://powerbi.microsoft.com/en/)
*   [selenium](https://www.selenium.dev/)
*   [pandas](https://pandas.pydata.org/)
*   [persiantools](https://pypi.org/project/persiantools/)
*   [deep-translator](https://pypi.org/project/deep-translator/)



Scraping and Dashboarding My Travel History from a Ride-Hailing Application

In this project, using the data I collected from Spotify (audio streaming and media services provider), I tried to make a recommendation system that based on audio feature, song genre, artist, etc., the song we choose gives us Suggest songs

[Project Github Repository](https://github.com/meysamraz/spotify-song-recommendation-app-end-to-end-project)

[Project Demo](https://spotify-song-recommendation.herokuapp.com/)

# Project Overview :

![](/images/viz1-8140de3c.jpg)

![](/images/viz2-5d2024dc.jpg)

## 1 - Setup Spotify Api

*   Open [spotify developers](https://developer.spotify.com/dashboard/login) go to ***dashboard*** - ***create app*** and get your ***client id*** and ***secret id***

\*\*\*\*\*\*![](/images/guide-be656949.jpg)

## 2 - Collecting Data

*   ### Collect songs between 2017 and 2022 (1000 per year)
    *   id : song id
    *   song name : name of the songs
    *   artist name : name of artist who sing the song
    *   artist genres : genres that artist sings
    *   album genres : genres of songs album
    *   release_date : release date of song
    *   song link : link of song on spotify
    *   image : cover image of song
    *   song duration : song duration
    *   song popularity : song popularity on spotify

*   ### Collect songs that i liked

*   ### Collect audio features of collected songs
    *   Danceability: Danceability describes how suitable a track is for dancing based on a combination of musical elements including tempo, rhythm stability, beat strength, and overall regularity. A value of 0.0 is least danceable and 1.0 is most danceable.
    *   Acousticness: A measure from 0.0 to 1.0 of whether the track is acoustic.
    *   Energy: Energy is a measure from 0.0 to 1.0 and represents a perceptual measure of intensity and activity. Typically, energetic tracks feel fast, loud, and noisy.
    *   Instrumentalness: Predicts whether a track contains no vocals. The closer the instrumentalness value is to 1.0, the greater likelihood the track contains no vocal content.
    *   Liveness: Detects the presence of an audience in the recording. Higher liveness values represent an increased probability that the track was performed live.
    *   Loudness: The overall loudness of a track in decibels (dB). Loudness values are averaged across the entire track. Values typical range between -60 and 0 db.
    *   Speechiness: Speechiness detects the presence of spoken words in a track. The more exclusively speech-like the recording (e.g. talk show, audio book, poetry), the closer to 1.0 the attribute value.
    *   Tempo: The overall estimated tempo of a track in beats per minute (BPM). In musical terminology, tempo is the speed or pace of a given piece and derives directly from the average beat duration.
    *   Valence: A measure from 0.0 to 1.0 describing the musical positiveness conveyed by a track. Tracks with high valence sound more positive (e.g. happy, cheerful, euphoric), while tracks with low valence sound more negative (e.g. sad, depressed, angry).

*   ### Collect audio features of liked songs

*   ### Merge songs and their audio features

*   ### Save data into a csv file

## 3 - Understand the data

*   Shape of the data
*   Check column dtypes
*   Check is there any null values
*   Check the correlation

## 4 - Cleaning the data

*   Drop "album genres" column
*   Convert artist genres to str

## 5 - Analyzing Data

*   Number of songs per year
*   Songs duration distribution and (min,avg,max)
*   Song popularity distribution and (min,avg,mode,max)
*   Artist genres wordcolud
*   Analyzing audio features

## 6 - Apply Dmensionality Reduction

*   Split audio features from data
*   Standard Scaling Audio Features
*   Apply PCA
*   Put PCA Output Into a Dataframe
*   Visualize reduced dimension Data

## 7 - Apply Clustering

*   Select the right value of k for clustring
*   Fit reduced dimension data into Kmeans
*   Visualize clusterd data (kemans output)

## 8 - Create Recommendation System

To build the recommendation system, I used audio features, songs, song genres, singers, etc. The more data the features are closer to the songs in question, the more data will be displayed.

## 9 - User Interface

To create my user interface, I used [streamlit](https://docs.streamlit.io/) formwork, a powerful formwork that allows me to create the desired user interface completely using Python.

## 10 - Deploy on server

I used [Heroku](https://www.heroku.com/) a cloud platform as a service which provide a free hosting to deploy my app on it. it's and amzaing platform gave me so much flexbilte to deploy my apps

## Libraries and FrameWorks used in the project

*   [streamlit](https://streamlit.io/)
*   [pandas](https://pandas.pydata.org/)
*   [numpy](https://numpy.org/)
*   [plotly](https://plotly.com/)
*   [matplotlib](https://matplotlib.org/)
*   [seaborn](https://seaborn.pydata.org/)
*   [termcolor](https://pypi.org/project/termcolor/)
*   [wordcloud](https://pypi.org/project/wordcloud/)
*   [yellowbrick](https://pypi.org/project/yellowbrick/)
*   [spotipy](https://spotipy.readthedocs.io/)
*   [scikit-learn](https://scikit-learn.org/)



Spotify Recommendation System App End To End Project

##### [Project Github repository](https://github.com/meysamraz/Google-Maps-Amsterdam-restaurants-analysis)

If you have watched the Dark series produced by Netflix, you know that understanding this series and the relationship between the characters can be a bit challenging :D . In this project, I scraped biographies of serial characters and used NLP and Graph Analysis to determine the relationship between serial characters and their importance.

![](/images/graph_chars.gif)

![](/images/comm_chars.gif)

## Project Overview :

### 1- Scraping characters biography

![](/images/bios-8b80016d.png)

### 2- Extracting .text Files and Combine Them All Together

### 3- Extracting Relationships from Characters Biography

### 4- Extracting Characters Names and Edging Based on Their Relationships

### 5- Ploting Relationships Graph

*   ### 🟣 Most Important

*   ### 🔴 Semi Important

*   ### 🟡 Moderate Important

*   ### 🟢 Secondary characters

*   ### 🔵 NPCs

![](/images/Screenshot%20(1621\).png)

### 6- Who Are The Top 10 Most Important Charecters in Tv Show

![](/images/plt1-f8246c5e.png)

### 7- Who Are The Top Top 10 Intermediarie Charecters in Dark

![](/images/plt3.png)

### 8- Who Are The Top 10 Communicator in Dark

![](/images/plt4-11d247c7.png)

### 9- Detecting Communities Between the Characters

![](/images/Screenshot%20(1623\).png)

## Libraries used in the project

*   [pandas](https://pandas.pydata.org/)

*   [matplotlib](https://matplotlib.org/)

*   [pyvis](https://pypi.org/project/pyvis/)

*   [selenium](https://www.selenium.dev/)

*   [spacy](https://spacy.io/)

*   [community](https://pypi.org/project/community/)

*   [selenium](https://www.selenium.dev/)

*   [networkx](https://networkx.org/)


Graph Analysis and Relationship Extraction from Dark Netflix Series

Digikala is a prominent Iranian e-commerce platform offering a wide range of products, from digital devices and mobiles to laptops, books, and clothing. In this project, my focus was on creating a specialized clothing search engine based on exact color preferences and clothing types. The project primarily centers on men's t-shirts and shirts to showcase the functionality of the system.

[Project Github Repository](https://github.com/meysamraz/digikala-clothing-search-by-color)

### Demo:

<https://meysamraz-cloth-color-search-digikala-project.streamlit.app/>

### Demo ***Heroku version***:

**Update:** Heroku may discontinue free hosting after November 28, 2022. If this link does not work, please use the link above.

<https://clothing-search-by-color.herokuapp.com/>

![](/images/demo_g.gif)

# **Project Preview:**

## **1. Data Collection:**

*   Data collection was achieved by extracting data from a hidden API provided by Digikala, a prominent Iranian e-commerce platform. This method was chosen for simplicity and effectiveness.

## **2. Data Preprocessing:**

*   Preprocessing involved several key steps:

    *   Dropping null values to ensure data quality.

    *   Creating a "Type of Cloth" column to categorize clothing items.

    *   Creating a "Color" column to represent the color of each item.

## **3. Image Color Detection:**

*   For image color detection, the K-Means clustering algorithm was employed. A threshold value and specific conditions were applied to accurately identify colors on various clothing items.

![](/images/color_detection-4dc01baf.png)

## **4. Front-End Development:**

*   The front-end of the project was built using Streamlit, a Python library for creating web applications. Streamlit offers a straightforward approach to web development entirely in Python. However, some limitations were encountered when customizing certain widgets, particularly in the color selection section.

## **5. Deployment:**

*   The project was made accessible online through Heroku, a cloud platform as a service. Heroku provided a flexible hosting solution for deploying the application.

## **Installation and Execution:**

*   To run the project, follow these steps:

    *   Install the necessary requirements using `pip install -r requirements.txt`.

    *   Execute the project with `streamlit run main.py`.

## **Libraries and Frameworks Used:**

*   Streamlit

*   Flask

*   Pandas

*   scikit-learn

*   Heroku

## **Future Updates:**

*   In future iterations of this project, I plan to implement the following enhancements:

    *   Expand the clothing categories to provide a more comprehensive search experience.

    *   Improve the accuracy and granularity of color detection .

    *   Enhance the user interface and interactivity to address Streamlit's limitations.

    *   Explore advanced image analysis techniques to extract more detailed clothing attributes.



Clothing Search by Color, Data Collecting, and Web Deployment

#### [Link of the project](https://www.kaggle.com/code/meisamraz/customer-segmentation-clustering-eda-report)

# Preview

In this project, I aim to extract meaningful segments from the customer data of a grocery firm. These segments will assist the company in gaining a better understanding of customer behavior and similarities, enabling them to optimize their products and marketing strategies.

# Clearilyfing Problem :

1.  **Business Objectives:**

    *   What are the specific goals and key performance indicators tied to customer segmentation?

2.  **Application Focus:**

    *   How will the segmentation findings be practically applied—product refinement, marketing enhancement, or both?

3.  **Product and Marketing Emphasis:**

    *   Are there specific products, promotions, or marketing areas to focus on in the segmentation analysis?

4.  **User Behavior:**

    *   Should the analysis prioritize current or historical user behavior, and are there specific behaviors of interest?

5.  **Demographics and Behavior:**

    *   Should the segmentation prioritize demographics, behavior, or a balanced mix of both?

6.  **Collaboration:**

    *   Is collaboration needed with other departments, such as marketing or product development?

# What data do we have to solve the problem ?



*   #### customer profile data

    *   ID: Customer's unique identifier

    *   Year\_Birth: Customer's birth year

    *   Education: Customer's education level

    *   Marital\_Status: Customer's marital status

    *   Income: Customer's yearly household income

    *   Kidhome: Number of children in customer's household

    *   Teenhome: Number of teenagers in customer's household

    *   Dt\_Customer: Date of customer's enrollment with the company

*   #### user behavoir data

    *   Recency: Number of days since customer's last purchase

    *   Complain: 1 if the customer complained in the last 2 years, 0 otherwise

*   #### marketing engagement data

    *   ### Products

        *   MntWines: Amount spent on wine in last 2 years

        *   MntFruits: Amount spent on fruits in last 2 years

        *   MntMeatProducts: Amount spent on meat in last 2 years

        *   MntFishProducts: Amount spent on fish in last 2 years

        *   MntSweetProducts: Amount spent on sweets in last 2 years

        *   MntGoldProds: Amount spent on gold in last 2 years

    *   ### Promotion

        *   NumDealsPurchases: Number of purchases made with a discount

        *   AcceptedCmp1: 1 if customer accepted the offer in the 1st campaign, 0 otherwise

        *   AcceptedCmp2: 1 if customer accepted the offer in the 2nd campaign, 0 otherwise

        *   AcceptedCmp3: 1 if customer accepted the offer in the 3rd campaign, 0 otherwise

        *   AcceptedCmp4: 1 if customer accepted the offer in the 4th campaign, 0 otherwise

        *   AcceptedCmp5: 1 if customer accepted the offer in the 5th campaign, 0 otherwise

        *   Response: 1 if customer accepted the offer in the last campaign, 0 otherwise

    *   ### Place

        *   NumWebPurchases: Number of purchases made through the company’s website

        *   NumCatalogPurchases: Number of purchases made using a catalogue

        *   NumStorePurchases: Number of purchases made directly in stores

        *   NumWebVisitsMonth: Number of visits to company’s website in the last month

## 2 - Feature Engineering and Data Cleaning

*   Drop null values

*   Convert Dt\_Customer dtype to date

*   Create Dt\_Collab column

*   Delete outlier (vampires in this case :D) datapoints based on Age column

*   Create Age Column (age of each customer in unti 2021)

*   Simplifying Education categories

*   Simplifying Marital\_Status.

*   Labeling categorial columns

*   Create **Children** Column (Number of children that customers have).

*   Create **FamilySize** column (Number of family members of customers)

*   Create **TotalSpent** column (The total spents on products (wine,fruits,meat,fish,sweets,gold) that the customer has made during last 2 years)

*   Delete outlie datapoints based on ***Income*** column

*   Create **TotalPromotions** column (The total promotions user acceped in last 2 years)

*   Crate **TotalPurchases** column (The total number of purchases made by the customer in last two years)

*   Create **CollabTime** column (How long has it been since user registration)

*   Drop unneeded columns

*   Check Correlation between variables

*   Standard Scaling Features

## 3 - Apply Dmensionality Reduction

*   Apply PCA

*   Put PCA Output Into a Dataframe

*   Visualize reduced dimension Data

## 4 - Apply Clustering

*   Select the right value of k for clustring

*   Fit reduced dimension data into Kmeans

*   Visualize clusterd data (kemans output)

## 5 - EDA clusters

*   Whats the total members of each cluster ?

*   What is the spent of the members of each cluster compared to their income ?

*   Members of which cluster accepted more offer in the campaigns ?

*   Whats the number of family members of customers in each cluster ?

*   Number of children of each cluster ?

*   Which way do the members of each cluster buy more ?

*   Whats the age situation in each cluster ?

*   Number of purchases in each cluster ?

*   What are each cluster interested in buying ?

*   How is marital status in each cluster ?

*   What is the status of each cluster in terms of education ?

*   Which cluster showed more interest in discount ?

## 6 - Profiling Customers & Summarizing Report

# Examples :

![](/images/Screenshot%20\(1057\)-86b33d42.png)![](/images/Screenshot%20\(1061\).png)![](/images/Screenshot%20\(1062\).png)![](/images/Screenshot%20\(1063\).png)

Sunset fishing is a beautiful time of day to relax and get in touch with nature

Customer Segmentation and Behavioral Analysis in a Retail Company

##### [Project Github repository](https://github.com/meysamraz/scraping-and-advanced-nlp-analysis-on-IMDB-TLOU-reviews)

The series TLOU has become very trendy these days and there have been many contradictions about the main actors of this series, **Pedro Pascal** and **Bella Ramsey**. In this project, I tried to understand the critics' feelings about the acting of these two actors based on the reviews of this series on IMDB

## Project Overview :

![](/images/plot6.png)![](/images/plot5-a4e7d591.png)

*   Scraping TLOU Reviews from IMDB

I used Beautifulsoup and Selenium to scrape data from TLOU review's page

*   Understanding the data
    *   Shape of the data
    *   Check column dtypes
    *   Check is there any null values
*   Data Preprocessing
    *   Convert Date Format to Datetime
    *   Convert Rating Format to Int
*   Feature Engineering
    *   Create Sentiment by Review Rating
*   Analyzing Reviews
    *   What Rating Did the Users Give to the Series?
    *   Whats the Min, AVG, Max of Ratings?
    *   What's the number of reviews according to episodes release date?
    *   Review Sentiments Based on Rating
    *   What's the Most Frequent Words in the Reviews?
*   Extract Reviews with Bella Ramsey and Pedro Pascal Mentions and Apply Sentiment Analysis on Them

## Libraries used in the project

*   [pandas](https://pandas.pydata.org/)
*   [matplotlib](https://matplotlib.org/)
*   [seaborn](https://seaborn.pydata.org/)
*   [wordcloud](https://pypi.org/project/wordcloud/)
*   [pytorch](https://pytorch.org/)
*   [transformers](https://huggingface.co/docs/transformers)
*   [spaCy](https://spacy.io/)
*   [selenium](https://python-visualization.github.io/folium/)
*   [beautifulsoup](https://beautiful-soup-4.readthedocs.io/en/latest/)



Scraping and Advanced EDA and NLP Analysis on IMDB "The Last of Us" Reviews

## Introduction

Since I didn't have access to a sports coach, I decided to make a coach :) My idea was to use pose detectoin to track my body movements during exercise to see if my movements are correct or not.

So If I do the exercises correctly, they are counted for me to reach my desired amount

## How it works :

For the project, it can truck three sports movements and count them if you perform the movement correctly (Side raises, Standing cruls, Squats)

<img src = "/images/sports-345a6679.jpeg" width ="660" />

### Track My Pose

I realized that there is an angle between the joints to perform sports movements, so we can use the angle between the joints to track the correct sports movements.

To get the angle of my joints, I needed a tool that can track my body and joints movements in real time. I used [Mediapipe](https://mediapipe.dev/) for this, an open-source cross-platform framework with solutions for live and streaming media. Developed by Google, Mediapipe allows me to recognize my pose in realtime with appropriate fps, and each of the joints in mediapipe has a unique id, which makes my work much easier.

![](/images/MediaPipe-pose-BlazePose-Topology.jpg)

### Calculate Between Angles

After extracting the points of the joints, I needed to calculate the angle between the joints, for this I used the [numpy](https://numpy.org/) library to solve linear algebra equations.

After acquiring the ability to calculate the angle between the joints, I needed to find the correct angle of the sports movements and since I did not have much experience in exercising, I was able to find the correct angle of the joints of the sports movements using this [YouTube video](https://www.youtube.com/watch?v=BNsKEG3hIzI\&t=76s) and I was able to Use some simple conditions to implement them.

![](/images/angle.png)

### Implement  User Interface

And now it was time to implement an environment that had a suitable user interface, although I had given opencv a visual form in realtime, but I wanted to use a web environment for this, so I used Streamlit, a very fast environment. Based on Python, which allowed me to implement my model in the fastest way in the web environment.

Working with streamlit was very easy for me because it was completely written in Python and had strong documentation and it didn't take me much time to learn it.

![](https://upload.wikimedia.org/wikipedia/commons/7/77/Streamlit-logo-primary-colormark-darktext.png)

## Setup and Run Project

### 1 - Clone project : -

Open [Project github repository](https://github.com/meysamraz/Smart_Exercise)

Download project

OR

Clone project

### 2 - Install requirement libraries and packages : -

`cd yourdir/Smart_Exercise-master`

Install packages by running :

`pip install -r requirements.txt`

### 3 - Run Project

`streamlit run home_page.py`

And open :

http://localhost:3000/


Smart Exercise with Streamlit And Mediapipe

***[Project Github repository](https://github.com/meysamraz/grocery_website_download_page_ab_testing)***

**In this use case I am working as an analyst for a large grossery chain. One of the goals our company has, it's to drive more customers to download thier mobile app and register for the loyalty program.**

One of the goals of the company discussed here is to get more users to download the mobile app and sign up for the customer loyalty program.

##### The company is making a link change to a button in the app store and asking if clickthrough rates will increase for the app download page?

## Explain methodology step-by-step :

*   ### What data do we have ?

    *   IP Address: Client's IP Address
    *   LoggedInFlag: Indicates whether the user has a profile on our website
    *   ServerID: Belonging to the Treatment and Control group, 1: Treatment , 2-3: Control
    *   VisitPageFlag: The status of the user clicking the download link

*   ### Whats the Duration Of the Test ?
    *   1 Weak

*   ### What's our unit of analysis ?
    *   IP Address: Client's IP Address

*   ### What's our metric to measure success here  ?
    *   Number clicked in download

*   ### The size of our treatment and control groups
    *   1/3 of the treatment units and
    *   2/3 of control units.

*   ### What's our MDE (Minimum Detectable Effect) ?
    *   MDE 1%

*   ### Apply hypothesis testing and check assumptions
    *   Check Normality & Homogeneity

    *   Apply tests (Shapiro, Levene Test, T-Test, Welch Test, Mann Whitney U Test)

    *   Evaluate the results

### Libraries and Packages :

*   [matplotlib](https://matplotlib.org/)
*   [seaborn](https://seaborn.pydata.org/#:~:text=Seaborn%20is%20a%20Python%20data,introductory%20notes%20or%20the%20paper.)
*   [pandas](https://pandas.pydata.org/)
*   [numpy](https://numpy.org/)
*   [scipy](https://scipy.org/)


My Projects