Photo: I took it by SM-A520F.

The purpose of this study is to predict the population of Berlin in 2035. Population is a very important variable for city regulations and planning. Estimation processes about the population of Berlin, the capital of Germany, in 2035 were created in line with the regression models. Afterwards, the results were compared over the R² values. In addition, the forecast graph was drawn over the LSTM Network.

Dataset is downloaded from macrotrends website. Dataset has 3 columns (Year, Population, Annual Change) and 87 rows with the header.

In this project, as stated in the title, results were obtained through three different methods. These methods are as respectively listed below:

1. Linear Regression

2. Polynomial Regression

3. Long Short-Term Memory (LSTM)

Linear regression population prediction in 2035:

3593846.83239776

Linear regression R² value:

0.6086274988056017

Polinomial regression (degree=2) population prediction in 2035:

3727080.32005322

Polinomial regression (degree=2) R² value:

0.7346609389955037

Polinomial regression (degree=4) population prediction in 2035:

3596823.26672745

Polinomial regression (degree=4) R² value:

0.825311510724274

RMS (Difference between actual population prediction and predicted population):

37224.69914068592

Process took 8.878643035888672 seconds.

Before running the code make sure that you have these libraries:

• pandas
• matplotlib
• seaborn
• time
• keras
• sklearn
• numpy

If you have something to say to me please contact me:

• Twitter: Doguilmak
• Mail address: [email protected]