Prachi Jadhav Email & Phone Number

Mumbai, Mh, In

Mumbai, Maharashtra, India

Mumbai, Maharashtra, India

Mumbai, Maharashtra, India

Sales Forecasting Model for New Product Launches Using Clustering and Time Series Analysis

Mumbai, Maharashtra, India

Product Portfolio Optimization Using Time Series Analysis

Pune, Maharashtra, India

Products Classification (Multilabel)Developed a multi-label classification model to automatically categorize products based on their features and attributes. Data was gathered and model was trained on a dataset of over 10,000 products with multiple labels per product and achieved accuracy of 90% on test set and hamming loss of 2%. Using NLTK library and regex library, data preprocessing and feature engineering was conducted to prepare the text data for classification including tokenization, stemming and TF-IDF vectorization, word2vec, tokenizer etc. Multi label classification techniques including One Vs Rest, Binary Relevance, Classifier Chains and Label Powerset Classifier were experimented. Experimented algorithm for base classifier were MultiNomial NB, Logistic Regression, SGD Classifier, naïve bayes and Support vector machine. Experimented with Deep Learning techniques like CNN, DNN, LSTM, glove-LSTM and finalized DNN for production.

Pune, Maharashtra, India

Product Image ClassificationBuilt an Image Classification model based on CNN, VGG16 and ResNET50 for an E - Commerce Platform for productcategorization/search and product recommendation. Data Augmentation: Using “ImageDatagenerator” class from Keras, data generator was setup to pre-process and augment the input images and then use them to generate batches of training, test and validation data for a deep learning model to be trained and evaluated. Built CNN model using sequential API from Keras. Model was compiled using Adam optimizer, binary cross entropy loss function and accuracy as a metric. Created two call-backs on CNN model intended to prevent over fitting and improve accuracy of model on unseen data. Evaluated to accuracy of a classification algorithm using confusion matrix. Achieved a Recall of 91% and F1 score of 91% on test data. Business Impact: This model helped client by automating the process of product classification based on product images. Model helped in customer acquisition and customer shopping experience by providing accurate search results and relevant product recommendation.

Pune, Maharashtra, India

Power Consumption Analysis and PredictionBuilt model to predict power consumption in a manufacturing plant using time series and LSTM. Data was collected for hourly power consumption data for last 20 years and used it to train a deep learning model. Also applied response variable design(RVD) to analyze power consumption patterns in a commercial manufacturing plant. Using RVD technique, established relationship between power consumption and various factors such as weather conditions, time of day and appliance usage. Pre processing, Feature transformation and Feature scaling was performed to prepare data for analysis and modelling. Performed EDA and data visualization on power consumption dataset to understand the distribution of mean of different power consumption features resampled over a month, week and day and hour.Input data preparation for LSTM modelling. LSTM model architecture to forecast power consumption based on historical data using Keras API. Evaluation metric RMSE was used coming around 0.59 and R2 score was 0.89. Business impact: Reduced power consumption upto 15% and yearly cost was reduced to 12%.

Pune, Maharashtra, India

Complementary Products Analysis using Apriori Algorithm, Association Rules, and Cluster AnalysisExperimented with K-Means Clustering to the RFM segments to identify more specific and sub groups of customers with similar purchase patterns.

Pune, Maharashtra, India

Valuable data generated from customer interactions, including support tickets, chat logs, emails, and phone call transcripts. Datasets building for usage in Data Science Projects. Projects: Customer Cohort Analysis, Sales Prediction, Loan Default PredictionBy providing insights into the frequency and severity of customer issues, help prioritize which problems the data science project should focus on solving. Feedback was provided on the relevance and accuracy of the model outputs, helping data scientists refine and adjust the models based on real-world customer interactions. Assist in testing the models developed in the project, identifying cases where the model's predictions were incorrect or not useful, providing insights that lead to further model tuning. Ensuring that the data used in the project is up-to-date, reflecting any changes in customer behavior or new issues that arise. Personalized customer support strategies, using insights from the data science project to tailor responses and solutions to individual customers. Support to the data science team on the nuances of customer interactions and common issues, ensuring that the models and analyses are grounded in real-world customer experiences. Ensure that the data used in the project complies with data privacy and security regulations, advising data scientists. Relayed insights and results from the data science project to other stakeholders within the company, helping to align the project’s outcomes with broader business objectives

Bachelor Of Engineering - Be, Production Engineering, Distinction

Activities and Societies: Elected as GENERAL SECRETARY of PRODUCTION ENG. DEPT.:- organizing dept. level, college level, and National.

Cbse, Pcm, Distinction