Publications by Selcuk Disci
Trend Forecasting Models and Seasonality with Time Series
Gasoline prices always is an issue in Turkey; because Turkish people love to drive where they would go but they complain about the prices anyway. I wanted to start digging for the last seven years’ prices and how they went. I have used unleaded gasoline 95 octane prices from Petrol Ofisi which is a fuel distribution company in Turkey. I arrange...
6914 sym R (4085 sym/22 pcs) 84 img
Testing the Correlation between Time Series Variables
In the previous article, we examined trends and seasonality in gasoline prices in Turkey. This time we will examine whether the gasoline prices are related to the variables that are thought to affect gasoline prices the most by the Turkish people. One of the variables is the Brent crude oil prices that are averaged monthly in dollars; the other i...
3478 sym R (1492 sym/7 pcs) 30 img
Turkey vs. Germany: COVID-19
In Turkey, some parts of society always compare Turkey to Germany and think that we are better than Germany for a lot of issues. The same applies to COVID-19 crisis management; is that reflects to true? We will use two variables for compared parameters; the number of daily new cases and daily new deaths.First, we will compare the mean of new cas...
3995 sym R (3420 sym/7 pcs) 72 img
Model Selection: Adjusted Coefficient of Determination-Variance Tradeoff
In my previous article, we analyzed the COVID-19 data of Turkey and selected the cubic model for predicting the spread of disease. In this article, we will show in detail why we selected the cubic model for prediction and see whether our decision was right or not. When we analyze the regression trend models we should consider overfitting and unde...
3321 sym R (1345 sym/5 pcs) 28 img
Time Series Forecasting: KNN vs. ARIMA
It is always hard to find a proper model to forecast time series data. One of the reasons is that models that use time-series data often expose to serial correlation. In this article, we will compare k nearest neighbor (KNN) regression which is a supervised machine learning method, with a more classical and stochastic process, autoregressive inte...
7392 sym R (2621 sym/10 pcs) 38 img
Approaches to Time Series Data with Weak Seasonality
In the previous article, we have tried to model the gold price in Turkey per gram. We will continue to do that to find the best fit for our data. When we chose the KNN and Arima model, we saw the traditional Arima model was much better than the KNN, which is a machine learning algorithm. This time we will try the regression model as a machine lea...
5079 sym R (1827 sym/5 pcs) 50 img
Bootstrapping Time Series for Gold Rush
Bootstrap aggregating (bagging), is a very useful averaging method to improve accuracy and avoids overfitting, in modeling the time series. It also helps stability so that we don’t have to do Box-Cox transformation to the data. Modeling time series data is difficult because the data are autocorrelated. In this case, moving block bootstrap (MBB)...
4146 sym R (2526 sym/4 pcs) 4 img
Backcast a Time Series for Covid-19 Truths
A couple of months ago, Turkey’s Health Minister announced that the positive cases showing no signs of illness were not included in the statistics. This statement made an earthquake effect in Turkey, and unfortunately, the articles about covid-19 I have wrote before came to nothing. The reason for this statement was the pressure of the Istanbul...
2416 sym R (2753 sym/4 pcs) 21 img
Time Series Forecasting with XGBoost and Feature Importance
Those who follow my articles know that trying to predict gold prices has become an obsession for me these days. And I am also wondering which factors affect the prices. For the gold prices per gram in Turkey, are told that two factors determine the results: USA prices per ounce and exchange rate for the dollar and the Turkish lira. Let’s check ...
4708 sym R (3027 sym/8 pcs) 26 img
Dynamic Regression (ARIMA) vs. XGBoost
In the previous article, we mentioned that we were going to compare dynamic regression with ARIMA errors and the xgboost. Before doing that, let’s talk about dynamic regression. Time series modeling, most of the time, uses past observations as predictor variables. But sometimes, we need external variables that affect the target variables. To in...
2713 sym R (2048 sym/3 pcs) 20 img