Python,Machine learning,Data Science,NLP

Hrushikesh Sahu

Data Science Trainee at AlmaBetter Sambalpur

****************@*****.*** 784-***-**** linkedin.com/in/hk-sahu github.com/hrushikeshsahu19 PROJECTS

Meru taxi trip time prediction

AlmaBetter Verified Project

12/2020 - 01/2021, Bangalore

Built a regression model using GBM, Decision tree regressor, and XGBoost models to predict taxi trip time in Delhi for a time period of six months.

Used a folium graph to visualize the pick-up and drop-off locations and used heatmaps, which were essential for EDA.

Applied feature engineering to obtain new features such as distance, speed, peak hours, busiest days and used Pearson correlation, VIF values to avoid multicollinearity in Linear Regression.

Applied Lasso and Ridge regularisation for optimizing the fit of the model and used Gridsearch CV for hyperparameter tuning, which resulted in R- square score of 0.71 on the test dataset. Marketing Campaign Effectiveness Prediction

AlmaBetter Verified Project

01/2021 - 02/2021, Bangalore

Developed a stacked model using Logistic meta-classifier on-base classifiers such as XGBoost, SVM, and Random Forests to predict whether a customer will start a Fixed-Deposit as a result of a marketing campaign. Obtained the ROC-AUC score of 91% on the test data.

Treated multivariate outliers using Isolation Forest and applied SMOTE boosting on normalized data to mitigate the problem of class imbalance.

Used SHAP values to determine the most important features contributing to purchase such as the number of calls made during the campaign, bank balance, personal loan, housing loan etc. News popularity prediction on social media

Almabetter Verified Project

02/2021 - 03/2021, Bangalore

Built a regression model to predict the popularity of news on social media platforms such as Facebook, LinkedIn, and Google Plus. Used tokenization, lemmatization, and pos-tagging and leveraged the concepts of SpaCy library to carry out text processing on the given dataset.

Applied TFIDF vectorizer along with PCA to reduce the complexity of the dataset and used Gradient Boosting Machine, Random forest, and XGBoost models to come up with the best working model on the test dataset for each of the three platforms respectively.

Carried out bias-variance tradeoff analysis to optimize the fit of the model and used Gridsearch CV for hyperparameter tuning, which helped in achieving R-square score of 0.72, 0.76, 0.78 respectively for each of the social media platforms.

TECH STACK

Languages

Python, C,JAVA

ML Frameworks

Scikit-learn, Keras, Pandas, Numpy, Seasons,

Matplotlib, spaCy, Keras, Pandas, OpenCV, NLTK,

Plotly

Platforms

Jupyter Notebook, Google Colab, Spyder, MS Office

Databases

SQL, Oracle

EDUCATION

B.TECH In Computer Science and

Engineering

Indira Gandhi Institute Of Technology,

Sarang

2021,

8.55/10.0

XII-Higher Secondary

Yuvodya Junior College, Balangir

2016,

77.16/100

X-Secondary

Govt. High School, Kuhibahal

2014,

75.33/100

RELEVANT COURSESWORKS

Machine Learning(AppliedAI) (2020)

KNN, SVM, Bagging, Random Forest, Naive Bayes, Boosting, GBDT, Xgboost, K-Means, PCA, LDA, NLP

Statistics for Data Science(Udemy) (2019)

Probability distribution, confidence interval, Hypothesis Testing, central limit theorem,Co-relation, Regression ACHIEVEMENTS

Awarded Scholarship for Higher Secondary

Education by State Government, 2016

1st in District Level Cricket Competition, 2013

INTERESTS

Music Cricket Swimming

Tags: Regression, XGBoost, Gradient boosting machine, MSE, R-square, Decision tree, VIF, homoscedasticity, multicollinearity, Gridsearch CV, feature engineering, Lasso, Ridge, Pearson correlation

Tags: PCA, Anomaly Detection, Feature engineering, Imbalanced dataset, Oversampling, Shapley Additive exPlanations (SHAP), Isolation Forest, SMOTE, Marketing Campaigns

Tags: Regression, NLP, spacy, PCA, Random forest, XGBoost, GBM, MSE, R-square, dimensionality reduction, TFIDF vectorizer, tokenization,nltk, feature engineering, Bias-Variance

Contact this candidate