Nowadays data science can generate business value from data analysis.
Data analysis field involves such operations as scrapping the data, its processing and providing useful conclusions and visualization. Using data analysis techniques it possible to figure out which feature influences the most on the product price, predict currency value, build the recommendation engine, detect and predict fraud incidents etc. Having your data or obtaining public information, many useful, unexpected and profitable results can be collected.
To turn them into business assets we combine data science methods with corresponding IT tools and frameworks.
We perform objects detection, tracking and recognition using high performance video and image processing.
Separately we have experience with:
We use machine learning methods with CNN (TensorFlow, Yolo, ImageNet) as well as specific frameworks like OpenCV, OpenFace to achieve optimal quality and performance.
We use modern technologies for text operating in a human like manner, giving the opportunity to generate and analyze text concerning the key words, text sentiment, emotion, structure, etc.
They used for following business tasks solving:
We use Scikit-Learn, NLTK, ElasticSearch, gensim.
Predictive analytics allows to utilize all the current and historical facts to produce predictions about future events. the application built with predictive analytics techniques can provide with probability of different decision profit, which is state of the art approach in nowadays time of making big amount of immediate decisions.
We have experience with prediction of:
We use Scikit-Learn, NumPy, Pandas, XGBoost and other technologies.
Video processing solution for players detection and tracking on soccer field with further statistics gathering.
The combination of various computer vision technologies was used to achieve the good accuracy and performance.
Platforms and Technologies: Python, OpenCV, YOLO.
CV verification implementation for international logistic company from Netherlands. It’s done through the relevance analysis between the text contained in CV and external sources (e.g. a press release, social networks).
Web service that uses data from ElasticSearch builds LSI model and MatrixSimilarity model on the top of it to search relevance between the documents was developed.
Platforms and Technologies: Python, Tornado, Scikit-Learn, NLTK, ElasticSearch, gensim
The solution which assists psychologist actions during physiognomy process. It extracts facial features from images.
Platforms and Technologies: Python, OpenFace.
The property management company goes to NYC market. They decided to do the aggressive marketing campaign and requested quality analyze of competitors property management services to find the potential customer not satisfied with the actual service level.
The solution uses a scikit learn script that trains a classifier from a data about the property management services characteristics.
As a result of the project, the group of potential customers was recognized with 4000% increased probability of the service provider change.
Platforms and Technologies: Python, Scikit-Learn, NumPy, Pandas, Predictive Analytics