Mobile Games A/B Test
Cookie Cats is an extremely popular mobile puzzle game developed by Tactile Entertainment. This is a classic connect 3 puzzle game where players must connect blocks of the same color to clear the board and win the level. As players progress through game levels, they occasionally encounter levels that force them to wait a considerable amount of time or make in-app purchases to proceed. In addition
The data we have contains information about 90,189 players that installed the game while the AB-test was running. The variables are:
userid - a unique number that identifies each player.
version - whether the player was put in the control group (gate_30 - a gate at level 30) or the group with the moved gate (gate_40 - a gate at level 40).
sum_gamerounds - the number of game rounds played by the playe
The Chitest result tells us that there is strong evidence that 7-day retention is higher when the gate is at level 30 than when it is at level 40. The conclusion is: If we want to keep high 7-day retention — we should not move the gate from level 30 to level 40.
You can view the Python code of this project on GitHub by clicking on one of the pictures.
House Price Forecasting Through a Multiple Linear Regression
The purpose of this project is to predict a house price of a unit area based on several house parameters. In this case, Multiple Linear Regression is the best option to create an accurate model.
I was given a dataset containing a piece of information about house price per unit area depending on several parameters such as house age, distance to the nearest MRT station, number of convenience stores,
Before creating a model I plot the heatmap showing how variables influence the target value. The greater the absolute value of the coefficient, the greater the influence of the variable on the desired value. Also, variables may be influenced by each other, for example, longitude and distance to the nearest MRT station. It may lead to the wrong model setup. Also, I checked the P-value of each varia
The trained model has r-squared 0.709 and helps to forecast house prices depending on their features.
You can view the Python code of this project on GitHub by clicking on one of the pictures.
Retention Forecasting Through a Classification Model
The purpose of this project is to predict the users who are most likely to fail an online training course based on data about their performance. To do this, I used 3 different classification models and compared their prediction accuracy and ROC-AUC.
After analyzing raw data I prepared a dataset containing information about users' first 3 days' behaviour with the target column whether they passed the course or not.
Using this dataset I trained Random forest, k Nearest Neighbor and XGBoost models and compared them by Accuracy score and ROC-AUC.
The best option in our case was the XGBoost model with an accuracy of 0.935 and ROC-AUC of 0.84.
You