%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\documentclass{beamer}
\mode<presentation>
{\usetheme{default}
 \usecolortheme{default}
 \usefonttheme{default}
 \setbeamertemplate{navigation symbols}{}
 \setbeamertemplate{caption}[numbered]}
\usepackage[english]{babel}
\usepackage[utf8x]{inputenc}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

\title{Real Estate Market Analysis -- Final project}
\author{Justyna Gajewska, Stanislav Vereshko and Jacek Wrzos}
\institute{Faculty of Informatics and Electronic Economy}
\date{27th May 2019}
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\begin{document}
\begin{frame}
  \titlepage
\end{frame}

\begin{frame}{Our goals}
\begin{itemize}
    \item To build a model to value properties on a secondary real estate market in 2014 for Poznań using historical data.
    \item To show our skills of using statistical packages to perform Real Estate Market Analysis.
    \item To show our skills of preparing professional presentation of the results.
\end{itemize}
\vskip 1cm
\end{frame}


\begin{frame}{A few words about data}
\begin{itemize}
\item We used the historical data - 1820 unique values - on real estate sales in Poznań in years 2008-2013.
\item Variables included in our research: transaction date, trans dwelling rooms, trans dwelling floor area, trans dwelling storey, trans dwelling spaces, trans dwelling spaces floor area, city district and detailed location.
\end{itemize}



\end{frame}


\begin{frame}{Basic Statistics}

Below we present basic statistics of quantitative variables.
\begin{table}
\centering
\begin{tabular}{c|c|c|c}
Variable & Average & Minimum & Maximum \\ \hline
Trans dwelling rooms & 3 & 1 & 7 \\
Trans dwelling floor area & 52,3 & 13 & 176 \\
Trans dwelling storey & 301 & -0,5 & 17 \\
Trans dwelling spaces & 1,15 & 1 & 6 \\
Trans dwelling spaces floor area & 7,04 & 0,7 & 112 \\
Longitude & 16,9 & 16,8 & 17,1 \\
Latitude & 52,4 & 52,3 & 52,5\\ \hline

\end{tabular}
\caption{\label{tab:variables}
Table of average quantitative variables.}
\end{table}

\end{frame}

\begin{frame}{Categorical Variables}
Below we present most common of categorical variables.
\begin{table}
\centering
\begin{tabular}{c|c}
Variable & Most common \\\hline
Poznań precinct & Św. Łazarz \\
Poznań district & Nowe miasto \\ \hline
\end{tabular}
\caption{\label{tab:variables1}
Table of most common categorical variables.}
\end{table}

\end{frame}

\begin{frame}{Gradient Boosting Machine}
To forecast future prices, we used the Gradient Boosting Machine method.\\

 Gradient boosting is a machine learning technique for regression and classification problems,
 which produces a prediction model in the form of an ensemble of weak prediction models,
 typically decision trees. It builds the model in a stage-wise fashion like other boosting methods do,
 and it generalizes them by allowing optimization of an arbitrary differentiable loss function.
\end{frame}


\begin{frame}{How it's work}
Like other boosting methods, gradient boosting combines weak "learners" into a single strong learner in an iterative fashion.
\includegraphics[]{algorithm-iterations.jpg}
\end{frame}

\begin{frame}{GBM in R}
%% we wrote a script in r based on the GBM method, which already with a small number of trees gave us a positive result of about 0.305
%so we were adding more and more trees
%until we reached the maximum computational level in the r cloud
%so we stopped at 4500 trees, which gave us a score of 0.29708
%although we could probably get even better results if it wasn't the R cloud computing limit.
\includegraphics[]{R.PNG}
\end{frame}
\begin{frame}{Different methods}

For prediction House Prices we also used method Regression Trees.\\

For this, we use the function rpart (), method "anova" and "poisson"\\

But method Gradient Boosting Machine gave the most accurate results and difference between predict and real prices was minimal.

\end{frame}
\begin{frame}{Score}
   \includegraphics[]{score1.PNG}
\end{frame}

\begin{frame}
\begin{center}
Thank you for the attention !
\end{center}
\end {frame}
\end{document}