QuantVector is a London-based Machine Learning Consulting firm experienced in applying Data Science and Machine Learning models to business problems.
We have successfully completed projects involving regression, classification, time series, natural language processing, data mining, deep learning and reinforcement learning.
Scoping & Architecture Design
First, we need to understand your problem better. Once we analyze the problem requirements, we will work closely together to prepare a road map, and split the project into specific tasks with specific deadlines.
Data Collection & Storage
We perform state-of-the-art scraping and have sound expertise in sourcing data: from the web, documents, databases and APIs. We also do crawling and build bots which automatically achieve our clients’ tasks.
Once data have been successfully collected and stored, we will perform an exploratory analysis phase to find patterns and correlations.
We run thousands of experiments in parallel to develop a machine learning model. A model is the core of a machine learning system – trained on historical data it can predict the future trends or understand the semantics of a text.
Each Client Is Important. Because we are a small company; we do not run a massive amount of projects as a large business do. This allows us to dedicate quality resources to each project. We pay close attention to each of our client’s at every stage of a project.
Highly Qualified Staff. Most of our developers hold Master’s Degrees either in Computer Science or Applied Maths. Also, we have several Ph.D. graduates on our team. We are working on research projects in AI, Machine Learning, and NLP field.
Expertise You Need. We are specialists with acquired years of expertise in specific industries like Security, Finances and Automotive and want to bring our in-depth focus to your project.
We Go the Extra Mile. We are committed to our clients’ satisfaction at all times. We provide project management, development services, and weekly meetings.
QuantVector has completed multiple projects for medium-sized companies in the USA & UK. The projects have covered various sections of the machine learning and data science model toolbox. Checkout below a sample from QuantVector’s project portfolio.
Quantitative Trading Algorithm Development The project’s aim was to develop, optimize and evaluate the performance of an intraday mean-reversion trading algorithm. Initially, our team proceeded with accuracy evaluation of historical stock data. Secondly, it proceeded with applying different clustering algorithms to group stocks with similar statistical behavior. Thirdly, it developed the trading algorithm and searched for alpha in different time frames. Lastly, the team proceeded with the integration of all different parts of the strategy and evaluated the algorithm’s performance.
Twitter Sentiment Analysis The project’s aim was to extract tweets of 15,000+ US stocks and use natural language processing combined with classification algorithms to look for potential relationships between a stock’s social sentiment and stock prices’ behavior. Initially, our team used Python’s Tweepy to access Twitter API and to download tweets for the 15,000+ US Stocks monitored. Secondly, Python’s NLTK was applied to convert pure downloaded tweets to a reliable social stock sentiment stream. Lastly, different classification models were applied to research for relationships between stocks price history and twitter feeds.
Financial Data Extraction The project’s aim was to extract historical data of Global Stocks & Bonds from different web sources. Python Scrapy was used to create a reliable, fast and powerful scraping algorithm which runs daily and successfully stores unique data to MySQL database. The database is used to store historical data of 50,000+ unique financial securities. Hence, the team proceeded with database optimization for fast data retrieval.
Customer Churn Reduction The project’s aim was to investigate how an insurance company can reduce customer churn rate. A dataset of 50,000 customers was analyzed to identify customer characteristics leading to increased probability account closure. The team developed an artificial neural network using Python’s Keras which produced a classification model with prediction accuracy of more than 95%. The company currently uses this model, to adapt its marketing efforts based on the customers’ characteristics.
Web Based Application Development A shipping company assigned to QuantVector the development of a web-based internal management system. After a thorough examination of the company’s operations, the team proceeded with the front-end and back-end development of the web based application. The system is currently used by the company’s on board and onshore stuff to streamline its technical operations, maintenance and monitoring of its fleet.