Our work in ML and AI has covered several domains like image analysis for wounds, stock predictions, user recognition for insurance (neuro networks and confusion matrix to check the accuracy of bot), sports club management, claim prediction, audio synthesizer, patient engagement solution, credit card and loan application and child abuse.
smartData developers expertise on Django and Flask and have good experience of working on data analysis and predictions in web development and digital enterprises applications for ERP, eCommerce marketplaces, edtech, and financial services.
We take pleasure in the work we do. Would you like to take a tour of our portfolio? Let’s check.
Bank Additional
The classification goal is to predict if the client will subscribe to a term deposit. The data is related to direct marketing campaigns of a Portuguese banking institution. The marketing campaigns are based on phone calls.
Petrol Consumption
ML describes methods that use statistical models, real-time analysis and algorithms to find patterns in data. This app detects the average petrol consumption based on entities like average income, tax, etc by using Random Forest.
Radar Signal
Here the targets were free electrons in the ionosphere and we trained the system on 4 entities of the dataset. Based on them it predicts for the new data, whether it, is showing good or bad evidence by returning the label ‘g’ for good or ‘b’ for bad.
Mobile Price Prediction
On the basis of mobile specifications like battery power, memory, screen, wifi, bluetooth, etc. we are predicting price range of mobile using SVM and SVC algorithms with sklearn. An accuracy of 80 % is achieved by the system.
Wine Label Prediction
Here, the dataset (a set of images of wine labels) was trained and associated on price, type, origin, and other notes from the web. Using machine learning and OpenCV we successfully predicted the labels of test images.
NLP for EHR
The purpose is to process different clinical records using NLTK/NLP to extract textual data and process it to find out predictions. Here, tesseract is used as OCR to convert pdf, images, docs to text and process it with NLP models to extract the required information.
Sentimental Analysis
This application predicts positive and negative sentiments in a given text using NLP. The model was trained with two labels used for classification, ‘positive’ and ‘negative’ on huge datasets of twitter. Algorithm- Naive Bayes.
Heart Disease Diagnosis
A model is trained over raw data from open source to predict if a person has heart disease. It has 4 levels of prediction based on severity level, approx. 14 different features were used to train the model. Use of prediction model with sklearn helped achieve accuracy up to 80%-85%.
