Nejc Znidar
Data scientist, statistician and financial modeler
Data research, software development, forecasting and creating automation tools. Practice expertise in big data and IT. I help businesses to learn how you can best use your data for your company.
Data scientist with 5+ years of broad-based experience in building data-intesnsive applications, overcoming complex architectural and scalability issues in diverse industries. Proficient in predictive modeling, data processing, data mining algorithms, as well as in scripting languages such as Python and R.
Work.
RECENT PROJECTS.
Click on a project to get more information about it.
Weighted liquid rank reputation algorithm coding
I helped translating Java code into Python code now readily available on Github.
Research work on reputation engines
Together with SingularityNET, I helped develop a weighted liquid rank reputation algorithm which aims at reducing recommendation fraud via experiments on simulated Amazon-like marketplaces. I coauthored several papers.
Testing predictability of stock returns using Fama-French approach
Testing if we could predict future returns from past; this work is in line with active research in finance. It follows Fama and French (1993) methodology.
Multi armed bandits example
Implementing Thomson sampling with multi-armed bandits and comparing it to constant sampling.
Fantasy sport tool
Fantasy basketball tool that scrapes data from nba.com and estimates points by player in next match. Made in Python.
Forecasting electricity prices
R script using various forecasting methods to best predict time series data of future electricity prices.
Critique of academic paper on research methods
Academic critique of approaches taken in scientific paper.
Bitcoin price smoothing
Time series smoothing of Bitcoin prices for time period 11-01-2013 and 31-12-2013.
Making a simulation of burning forest in R
Simulation of a fire in the forest based on wind speed, wind direction, dryness and other relevant factors. All those factors are taken into account and binded in actual simulation.
Skills and experience.
PREDICTING THE FUTURE TODAY.
Having studied financial mathematics (Bsc.) at University of Ljubljana and Quantitative finance on Tilburg university (Msc.), my skills are mainly IT and mathematics. I also had an exchange year at Humboldt Universtiy.
Experience:
- UMTR trade group LTD. Period: July 2020-today.
-
Implemented a machine learning models to predict customer behavior. This helped assess marketing campaings better and faster and therfore optimize marketing spend.
- Implemented a machine learning models to predict customer behavior. This helped assess marketing campaings better and faster and therefore optimize marketing spend.
- Set up the model that predicts the success of specific marketing campaing with 80%+ efficiency.
- Reduced the company’s ad spend by 40% due to integration of data-driven approach to the marketing.
- Set up a system that allowed the company to execute data-driven decisions in their online shop department.
- SingularityNET. Period: September 2017-June 2020.
- Created a stochastic financial model, which quantified risks for investors prior to successful fund raising.
- Developed of reputation system called liquid rank reputation system for online marketplaces. A system helps recommending products and minimizes fraud.
- Made a research exploring cause and effect relations in climate niche.
- Zalando SE. Tamara-Danz-Straße 1, 10243 Berlin, Germany. Period: July 2016 - Februar 2018.
- Performed statistical analysis using different regression and time series model which then drove data-driven decisions regarding ad expenditure.
- Helping with analytical data and reports, analysis and forecasting to support SEO, SEM, affiliate, gift voucher and other decisions.
- Forecasted and estimated ad performance using statistical and machine learning algorithms.
- Made the model which optimized and helped automize ad spend.
- CGI research, period: September 2018 - February 2019
- Set up a trading algorithm in cryptocurrency niche.
- Novamente llc,
- Actuarial practice at Triglav d.d. Miklošičeva cesta 19, 1000 Ljubljana. Period: September 2013.
- Basic actuarial tasks, examination of new program code, examination of online insurances and examining competitiveness.
- Head of human resources AEGEE - non profit. Oct 2013 - Oct 2014.
- Was in the board of Aegee Ljubljana 2013-2014. My role: Human resources. We organised several events in Ljubljana and other Slovenian towns and summer university in cooperation with Aegee Udine.
- Foreman assistant in Cestno gradbeno podjetje Novo mesto (CGP) (construction company). Period: June and July 2009 and July, August 2010.
- Assistance at accounting, quotation creation, terrain work and others.
List of my skills: Mathematics, SEO, econometrics, statistic, programming.
Statistical models: random forests, probit, logit, decision and regression trees, ARIMA, ARCH models, multilayer perceptron, causality detection and measurement, etc.
Computer skills: Basic program work (Office (including products such as PowerBI, Pivot tables, etc.), basics of HTML, LateX,…) and programming (in AutoIT (self-learned), Python, R, SQL and Matlab). Also Microstrategy, Business objects, Tableau.
Education
- Bsc. in financial mathematics at University of Ljubljana (math faculty). Time period: 1st October 2012 - 8th September 2014.
- Msc. in Quantitative finance at Tilburg University (located in the Netherlands). Time period: 1st September 2014 - May 2016.
- Exchange student, Humboldt University (Berlin). Time period: 1st October 2015-31st March 2016.
Recent projects
- Worked in Zalando 1 year 8 months. Worked on statistical data driven models to optimize marketing performance of a company. The goal was to optimize spending and get as many profitable customers as possible. Moreover, trying to quantify the exact effect of different marketing campaigns. Programming languages used: R, Python, SQL.
- SingularityNET: Built a stochastic model to predict possible future movements of cryptocurrency prices; used for risk management and decision steering. Programmed in Python.
- Causality detection: Determining causality in neuroscientific research; did EEG wave scans react to given signal? Programmed in R.
- Building fantasy basketball tool which forecasts most likely amount of statistics per player based on their past performance. Data was scraped from nba.com using APIs. Programmed in Python.
- Make a model to forecast future electricity prices based on past data. Time series analysis and forecast done in R.
- Writing a critical review of different statistical methods chosen in scientific paper (mainly what could be improved to make conclusions stronger).
- Writing a code to track cryptocurrencies in Matlab together with different crypto index choices. And then translating this code into R.
- Made a predictive model to predict companies' chance of going laying off people based on past layoffs, stock price movements, etc. to get an early automated signal. Done in R.
- Replicating a scientific study which calculates market premium using Ross recovery theorem (get's current state prices from real world option data). Done in Matlab.
- Made a script connecting BigQuery and Adwords to get some data engineering tasks done that. Using APIs and Python.
- Created predictive model to forecast future sales based on past data, seasonality and marketing efforts. Using Python.
- Simulating the burning forest given the wind directions, forest distribution, etc. Programmed in R.
- Creating R shiny application that determines the success of an add on SEA based on content written in description, headlines, path taken, etc.
- Teaching R (basics and some statistical approaches) 40+ people.
Education
- Bsc. in financial mathematics at University of Ljubljana (math faculty). Time period: 1st October 2012 - 8th September 2014.
- Msc. in Quantitative finance at Tilburg University (located in the Netherlands). Time period: 1st September 2014 - May 2016.
- Exchange student, Humboldt University (Berlin). Time period: 1st October 2015-31st March 2016.
Services.
01.Forecasting and automation
I am building forecasting models that help bussinesses automate processes and better expect the future cash flows. My skills combine software development as well as statistics and machine learning. Models that I've worked on in the past are:
- Logit and probit models (multinomial, fixed and ordered).
- Fixed and random effects model.
- Polynomial regression.
- Time series models such as GARCH ARIMA, ARMA and moving average.
- Decision trees such as CART, random forest, boosted trees.
- Multilayer preceptron
- Multi-armed bandits - Thompson method.
- Ford-Fulkerson algorithm.
- Linear programming.
- Gradient descent, Newton method.
- Quasy-Newton methods.
- Lebesgue integral and Riemann integral.
- Radon-Nikodym theorem.
- Linear regression and generalized linear model (GLM).
- Natural language processing
The use of statistical analysis and machine learning methods already helps to save immense costs, to optimize processes and to predict the expected sales.
Good examples include logistics optimization, reduction of machine failures through permanent data stream monitoring and the reduction of marketing costs through more targeted customer acquisition, automating processes online with recommendation systems, user personalization, self driving cars, etc. My aim to provide you with permanent tools to the highest standards, which contribute to the long-term security of your company's success.
It goes without saying that the contents of our discussions and projects are kept strictly confidential.
My personal goal is to realize examples like these in small and medium-sized businesses. Often, it surprising how some good data handling can positively surprise you.
02. Financial modeling
I am making financial models for different purposes. They can be used for risk analysis, risk management, trading or as tools for any other financially related application. I can help with financial and insurance related problems.
I can help to construct a financial representation of some, or all, aspects of the firm or given security. My focus is in quantitative finance applications. I can help you determining prices of derivatives and other financial instruments. I am building tools and help analyze the models and financial movements. Moreover, I can help you with stochastical analysis and modeling of financial markets.
Having some background in game theory and stochastic calculus helps me understand some reasons for investor and human behavior. I can help with following things:
- Partial differential equations.
- Stochastic differential equations.
- Derivative pricing; Black-Scholes model, Vasicek model, Hull-White model, Cox–Ingersoll–Ross model, Chen model and some others.
- Game theory; Bayes games, cooperative games, bimatrix games, strategic games and some others.
- VaR calculation (value at risk).
- Monte Carlo method.
- Fourier transformation.
- Numerical analysis.
- Homogenic Poisson processes.
- Martingales, stopping times.
- Hedging and replication.
- Contigent claim pricing.
- Swaption pricing.
- Annuity pricing.
- Pricing of life insurance products.
- Sharp ratio, hazard rate and Wang's premium principle.
I can help you with models that will satisfy regulatory requirements as well as make models for your business that concern its valuation of long term financial projections. I can also help you with pricing derivatives and determining prices of different instruments.
03. Data analytics
Data analytics is a process of inspecting, cleansing, transforming, and modeling data with the goal of discovering useful information, suggesting conclusions, and supporting decision-making. I analyze dataset on continuos basis and can help you with basic and not so basic modeling of the data.
My focus in this area is more on analyzing data, mostly doing data mining, however I also do quite a bit of business intelligence tasks. Furthermore, I use PoverPivot in excel, have some experience with PowerBI and can also work with Tableau for visualizations.
In this area, I can help you with the following things:
- Descriptive statistics; explaining the data using graphical representations such as scatterplots, correlation metrics and conditional distributions.
- Exploratory data analyzes, by using boxplots, histograms, odds ratios, etc.
- Statistical hypothesis testing. I worked with many different testers. I will only mention a few here: Jarque-Bera test, Kolmogorov–Smirnov test, Pearson's chi-squared test, Diehard tests and others.
- Quantiles and percetiles of distribution.
- Causality detection - using tests like Granger causality or multivariate Granger causality.
- Causality determination - using for example causal impacts (Bayesian structural time-series models).
- Using other well-known statistical measures such as deviation, median, mean, etc.
I can help you with statistical analyses for your business. Furthermore, interesting thing to understand here is causality. Correlation does not imply causation, as it is often said. I can help determine the difference between the two using more advanced methods to actually analyze one or the other. Various analytics usually need to be done for different datasets and I can help you with those as well.
Clients.
MAKE A DIFFERENCE.
“ Nejc is a great guy to work with, He is an expert at Data Science with great collaborative skills. Worked with him on a month long project analyzing huge data sets by creating learning and testing models for machine learning. He exceeded my expectation and will absolutely work with him again. ”
Victor M.
“ Nejc is fast, knowledgeable, careful and adaptive... capable of doing complex tasks based on high-level instructions and asking questions as needed. Very professional at writing up results as well. ”
Ben Goertzel, founder of SingularityNET, Novamente LLC, OpenCog and several other endeavours