Adil Salim

Well-qualified Data Scientist experienced working with vast data sets to break down information, gather relevant points and solve advanced business problems. Skilled in predictive modeling, data mining and hypothetical testing. Offering more than 2 years of experience in data visualizations and model building.

Current Address

'Anmol', Othungode, thirunellai (PO), Palakkad, Kerala, 678004

Passport Number

M9494739

• Awarded Gem award for the month of February (2017) for the highest quality score in McKinsey Trivandrum
• Rewarded with promotion to grade 2 within 10 months of hiring
• Completed a Certification course on Tableau10 from Udemy.
• Certification of Appreciation from Government of Kerala for clean-up operations at the Sabarimala
  temple and Pamba river
• Web application for Market basket analysis and sales prediction accepted by Bagetid.dk (the online arm of Dr. Oetkers) as a solution for increasing theor online sales
  
  

I have worked as a freelancer for many individual clients requiring services in analytics and visualizations through various applications such as fiver and freelancer.

Sentiment Analysis of Airbnb User reviews using RapidMiner

• Objective: To find the general sentiments of customers regarding each property and give them a sentiment score based on the relative sentiments of satisfaction or dissatisfaction in comparison to each other
• Abstract: From the more than 16,000 properties in the database I have filtered the ones with user reviews with comments, these were then run through the sentiment analysis operator provided by Aylien and the scores have then been averaged out for each of the properties and the final sentiment scores have been given.
• Tools and Skills: RapidMiner; Text analysis, descriptive analysis and sentiment analysis
• Results: Created a model which when cross-verified with the ratings showed a high level of prediction accuracy

Market Basket Analysis and Sales Forecasting for Bagetid.dk (online presence of Dr.Oetker)

• Objective: To find the products that were frequently bought with one another and the products that drove the sales of other products (Locomotives and Wagons)
• Abstract: Bagetid.dk provided me with the sales data from June of 2018 till June 2020 of all their listed products from these I have converted them to orders and then found the frequently bought pairs using apriori algorithm and then used fbprophet to predict the sales of individual products. I have then made a web application using Streamlit to allow the marketing manager to automatically find the frequently bought pairs for each product and the predicted sales ( with all prediction variables changeable based on interest)
• Tools and Skills: Python; Streamlit, FBprophet, apriori algorithm, market basket analysis and time series forecasting
• Results: The final solution was greatly appreciated and put into use by bagetid.dk

Deep Learning for Image classification using TensorFlow

• Objective: To build an image classification model using TensorFlow
• Abstract: The given dataset had more than 40,000 images of vehicles with 10 possible types of vehicle and with a classification id to show that they belong to a specific class. I have used the multiple layers and classification networks to classify the vehicles with a high level of accuracy
• Tools and Skills: Python; TensorFlow, deep learning and image classification
• Results: The model was able to predict with an accuracy higher than 90% for even unclassified images and external images of vehicles

Choosing preferable mode of transport by Employees

• Objective: To build a model for deciding on the mode of transport that the employees prefer while
  commuting to office
• Abstract: For this, multiple models such as KNN, Naive Bayes, and Logistic Regression have been
  created and explored to check their model performance metrics. Bagging and boosting modelling
  procedures have also been applied to create the models and model predicted mode of transport was
  provided with justifications
• Tools and Skills: Python; Bagging and Boosting, KNN, Naive Bayes, Logistic Regression
• Results: A prediction model of higher than 90% accuracy and kappa with validation

Project Title: Building a supervised Model to cross-sell personal loans

• Objective: To build a model using a Supervised learning technique to figure out profitable segments
  to target for cross-selling personal loans
• Abstract: A Pilot campaign data of 20000 customers was used which included several demographic
  and behavioral variables. The Model was further validated and a deployment strategy was
  recommended
• Tools and Skills: Random Forest, Data Mining, Pruning, Model Performance Measures
• Results: A prediction model of higher than 90% accuracy and kappa with validation
  
  
• Tools and Skills: Python, Text analysis, descriptive analysis and sentiment analysis