A/B testing, also known as Split Testing, is the technique of determining if there is a statistically significant difference in the outcomes of control and test groups. But the two groups might not be normal and might not be homogeneous, hence we have to test for normality and homogenity first, then apply a suitable statistical test.
The final policy is used to make the agent, which like the analytical agent, follows some rules.
The idea behind Double Regression is to perform multiple and separate regressions on the effect of Confounders and Instrumental variable (e401) on Target variable (net_tfa), the effect of Confounders and Instrumental variable on Control variable (p401), and the effects of Confounders on Instrumental variable.
I made the class for the environment using OpenAI gymnasium and initialized an environment.
Here are some example applications of the indicator at work, giving buy and sell signals:
This library works best with hyperparameter tuning. Slightly different parameters can give significantly different results.
This library is available on git. It can be installed with
Introduction
Causal Analysis is the study of causation as opposed to correlation. Given two events A and B which appear to be correlated, can we determine with a certain statistical significance if one of them is the cause of another?
This section details the skills I have acquired, projects I have done and papers I have published without any academic or professional incentive or aid.
These are the libraries and packages I have developed to automate my workflow and encapsulate my algorithms.
This is a library encapsulating the algorithm developed by me for time series forecasting in general. It has been used for Inventory Management (particularly Sales Forecasting), Liquidation Forecasting and numerous other projects of the Syngenta INPU data science intern team.
I learnt Algorithmic Trading from the NPTEL course Algorithmic Trading and Trading View.
I was introduced to Quantitative Finance through the project Personal Finance.
I used to be fascinated about Quantum Computing and Quantum Mechanics in general in my schooldays, hence deciding to do an Integrated MS with Physics major at Indian Institute of Science, Bengaluru.
This is the Kaggle competition regarding Game AI and Reinforcement Learning.
Reinforcement Learning was something that deeply interested me. I was fascinated by the idea of an AI that can learn from experience.
My first full-time internship was in the supply chain department of Syngenta so I naturally learned about it and did some projects.
This project was started as part of an intra-company (Syngenta) hackathon but I continued doing it as a personal project after the hackathon's completion.
The library has the following parameters. Some of them have default values but its better to tune them to optimal values for best results.
The policy is a custom CNN policy made using PyTorch:
The first step is to detect the current trend, i.e., bullish or bearish or sideways market. Buy signals will always be generated at the bottom of a bearish trend and sell signals at the peak of a bullish trend.
Naturally, the bottom of a bearish trend and peak of a bullish trend is when the trend reverses. To detect trend reversal, I used a modified version of RSI.
The first problem is to find the optimal packaging of products in vehicles (trucks in this example) so that
This library can be used in two different ways. But in both ways, you have to perform these steps first:
Next step is finding the optimal route each truck (vehicle) should take to minimize transportation cost (length of route) while making all deliveries on time.