I can help you to..

---

• Harvest data from different sources like APIs, SQL databases or messy Excel spreadsheets.
• Reformat and reshape data to make further analyses easier.
• Extract and clean impactful data from websites.
• Clean datasets removing outliers and abberant data.
• Create reproducible data preparation workflows
• Automatize the creation of data related reports

Example:

---

• I collected 1000+ Data scientist Salary Data from the Glassdoor.com to analyze how different locations affects data scientists salaries and to make a predictive model API for people to use. I used the Selenium module in python to extract and clean the data. and used Flask to create an endpoint for the model
• Developed a simple GUI based application that scraps all the property detail from zameen.com, and Apkaghar.com, cleaned and transformed the data and then inserted that into the database.

Contact me

---

Tell me more about your project at zainulabedin.marwat@gmail.com to see if we can work together.

I can help you for

---

• Exploratory analysis of a dataset.
• Creation of clean reports using Jupyter Notebook.
• Extract meaningful insights from your dataset.
• Application of statistical methods on your dataset.
• Creation of custom scripts for analysing your data automatically

Example:

---

• I created an interactive realtime Pakistan district Wise Map for a department called PMRU civil secretariate. It is directly plugged on their SQL database and allows to access its information in real time presenting meaningful information/grapphs on mouse hover.
• I analyzed All KPK(Pakistan province) schools and colleges data to improve quality of education and to reduce overall cost.

Contact me

---

Tell me more about your project at zainulabedin.marwat@gmail.com to see if we can work together.

I can help you with..

---

• Creation and deployment of Machine Learning Model.
• Training and retraining ML systems and models as needed
• Extracting, Cleaning and Transforming data for the Machine Learning algorithms.
• Creation of Deep learning models using numpy and tensorflow..
• Creation of pipline from data cleaing to prediction.

Example:

---

• Built a ML pipeline that can differentiate between 10 types of tomato leaves, the model used was VGG19 model.
• Worked on association rule mining recommender system, that suggest an item that a person may like based on frequent sets of items that are also bought along with it, the algorithm that was used is Weighted Apriori Algorithm

Contact me

---

Tell me more about your project at zainulabedin.marwat@gmail.com to see if we can work together.

---

Every penny that I earn, 10 to 30 percent of it goes to charity. I hand pick housed and people in my country that can not afford monthly wages and give them accordingly. The percentage fluctuation depends mainly on the type and longevity of the project.
It can also be your choice of charitable organization that my donations (you can also decide the percentage(10-30) of it) can go to whenever you Hire Me.

Contact me

---

Tell me more about your project at zainulabedin.marwat@gmail.com to see if we can work together.

Description

---

As Immigrant myself, I have been keen to know which countries migrates the most. Upon research and I found a dataset, that contains immigrants from different countries to Canada.

I built a heat-map which indicates each countries migration to Canada, the darker the area/country the more migrants they had. I used Python, Pandas, and Folium tools to create this beautiful visualization.

Read more

---

If you want to analyse your geographical data for useful insights, you can get in touch with me through: zainulabedin.marwat@gmail.com

Description

---

Crimes have been happening since dawn of time, though, knowing which area to avoid or keep an eye out for might help some in the future. This map let you see all the 150500 crimes that happened in San Francico.

I have created an interactive map, which have clustered all the crimes into different area, as you zoom in/click on crimes they are divided into many other clusters and soon till there are only few clustered/non-cluster crimes left. The tool was created using Python and Folium, plus several Html widgets for interactivity.

The below figure shows the clustered crimes all over the San Francisco

---

When zoom in on it, you can see the following.

---

Read more

---

The Html file of the project can be found on Github. Due to privacy reasons I cannot share the source code or the data.

Description

---

As a Data Scientist practitioner, it would really comes in handy if I had a system that could suggest a salary based on some metric. To do this I scraped thousand of data science jobs description/salary and much more from Glassdoor.com, trained a model and builit an API to use in the future for predicting DS salaries with MAE of ~11K


I trained and tuned differnt models, and selected the best performing machine learning model, and created an API for people to integrate into their systems. I used Python, scikitLearn, pandas, statsmodels and pickle, matplotlib for pre-processing, EDA and training the model, and used Flask to create an endpoint.

Sector-wise Data Scientist Jobs

---

Read more

---

The source code of the project and the dataset are availableon Github.

Description

---

PMRU is a government company with the aim to improve the performance of all tiers of provincial government through innovative use of Information and Communication Technology.

I created an interactive district-wise geographical map allowing to explore each district performance. It allows to quickly answer following questions.

• How well is each district governmental departments performance?
• What are the most important concerns of the people of each district?
• Does performance of the government departments increases or decreases quaterly?

This way government can take informed steps to make better service delivery to the citizen at local level.

All of the above information is gathered from a very large Government SQL Database.

---

Below graphs shows just a portion of the map and data when mouse is hover over a district.

---

Due to privacy reasons, I cannot show other sensetive information

Read more

---

I can't provide more information on this project for privacy reasons.

Description

---

Businesses that tends to their customers reviews are the businesses that are most likely to succeed. Every business have customers feedback and they try to tend to them as quickly as possible, though, with thousands and thousands of reviews/feedbacks it is very hard to differentiate complaints or questions type of reviews which takes time and time is of essence for business.

To tackle that I have created a web app which takes an .csv file which have one column with reviews the app then use Machine Learning model which has been trained on millions of reviews from obtained from amazan and classifies ther reviews into 5 categories, label them and download the file for users to use.


I have used Django as the framework for the application and used HTML, CSS, JavaScript for dynamic web application and used many other crucials modules.

Read more

---

The source code of the project can be found in my github repo.

Description

---

To Institue of Marince Research Norway it was of very importance to estimate the population of seals, particularly seal pups.

I was approached by a researcher to help him with creating a classifier that could classify harp seal pups from hooded seal pups.

I achieved this by first using CNN then reducing the feature size by applying PCA, then I used Logistic regression to classify the seals. The model was able to achieve 88.26% accuracy, with AUC as given below.

AUC score

---

Read more

---

Due to privacy reasons, I cannot share the data or the source code