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

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

In this project, I scraped details of Amsterdam restaurants from Google Maps using Selenium and analyzed the data.

Who is this analysis for ?

*   Tourists, those who travel to Amsterdam and those who live in Amsterdam can use this analysis to find the best restaurants.
*   Those who are looking to start a business in the restaurant industry and provide services in Amsterdam

I used Google Maps to get data about restaurants in Amsterdam. Unfortunately, it is not possible for those who live in Iran to access Google APIs :( But this is not the end of the work, I was able to create two robots using Selenium and scrape the Google maps data. The first bot [Google_maps_scraper.py](https://github.com/meysamraz/Google-Maps-Amsterdam-restaurants-analysis/blob/master/Google_maps_scraper.py) is from the Google Map site for scraping. It uses and is able to scrape restaurant reviews and the second bot [Google_scraper.py](https://github.com/meysamraz/Google-Maps-Amsterdam-restaurants-analysis/blob/master/Google_scraper.py) uses Google to collect data.

# Project Overview :

![](/images/demo_scraper1.gif)![](/images/demo_scraper2.gif)![](/images/plot3-6cf1add7.png)![](/images/plot4-74c0a95e.png)![](/images/plot5-bcbb0dc0.png)

*   ### 1 - Scraping data from Google maps
    *   title : name of restaurant
    *   rating : google maps rating (star 0-5) of restaurant
    *   reviews : number of reviews of the restaurant
    *   expense : price range of the restaurant ($ - econony, $$ - regular, $$$  - expensive, $$$ - luxury)
    *   category : what type of food does restaurant serve (Italian - french or etc.)
    *   description : restaurant description in Google maps
    *   services : What kind of services does the restaurant offer? (die-in,takeaway...)
    *   address : The registered address of the restaurant in Google Map

*   ### 2 - Cleaning the data
    *   Convert ratings to number
    *   Remove K and convert number of reveiws to int
    *   Remove "Service options:" from services

*   ### 3 - Analyzing Data
    *   The Number of scrapped restaurants
    *   Min - Meam - Max of Restaurats Rating
    *   Rating vs Reviews
    *   Rating Distribution
    *   Removing outliers and perform analysis
    *   Min - Meam - Max of Restaurats Number of Reviews
    *   How many restaurants have takeaway?
    *   How many restaurants have dine-in?
    *   How many restaurants have both dine-in and takeaway?
    *   Restaurant with most reviews
    *   Restaurant with lowest rating
    *   Restaurant with highest rating and No. reviews
    *   What categories most restaurants are from?
    *   Number of closed Restaurants
    *   In what category are the restaurants in terms of price?
    *   Restaurant with highest rate in each category
    *   Reviews Word Cloud

*   ### 4 - Get latitude and longitude based on the addresses
    *   Which area of Amsterdam has the most restaurants?
    *   Where are most luxury restaurants located ?

## Libraries used in the project

*   [pandas](https://pandas.pydata.org/)
*   [numpy](https://numpy.org/)
*   [matplotlib](https://matplotlib.org/)
*   [seaborn](https://seaborn.pydata.org/)
*   [termcolor](https://pypi.org/project/termcolor/)
*   [wordcloud](https://pypi.org/project/wordcloud/)
*   [geopy](https://geopy.readthedocs.io/)
*   [folium](https://python-visualization.github.io/folium/)



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

***[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/)


Grocery Website Download Page AB Testing

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 ?

Whats the Duration Of the Test ?

What's our unit of analysis ?

What's our metric to measure success here ?

The size of our treatment and control groups

What's our MDE (Minimum Detectable Effect) ?

Apply hypothesis testing and check assumptions

Libraries and Packages :

Read next

Scraping & EDA Amsterdam Restaurants and Cafés on Google Maps

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

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