\documentclass[11pt]{article}

\usepackage{graphicx}
\usepackage{multirow}


\begin{document}


\begin{table}[ht]
\caption{Best result for each category}
\begin{center}
\resizebox{\textwidth}{!}{\begin{tabular}{p{0.7cm}p{2cm}p{2cm}p{3.5cm}p{1.5cm}p{2cm}p{3.5cm}}
	\hline
	\multirow{2}{*}{Num}& \multirow{2}{*}{Topic}& \multirow{2}{*}{Articles}&  \multirow{2}{*}{Techniques} & \multicolumn{3}{c}{Best Result}\\
	\multirow{2}{*}{}& \multirow{2}{*}{}& \multirow{2}{*}{} &  \multirow{2}{*}{}& Article & Method & Result\\
	\hline
	X&X&X&X&X&X&X\\
1& Forecasting Cash Demand & Article.No 1,2,5,6,7,8,11,14,15, 
16,17,18,19,20,
27,29,40,43,44 & 
Time series ,Neural Network, KNN, LWL-based model, ANN, SVM, Genetic Algorithm, Fuzzy Wavelet-based, SVR, Feed-Forward Neural Network, LS-SVM, MLP &	No.6 \cite{andrawis2011forecast} &	Time series methods &	18.95\%  SMAPE \\
\hline
2&	ATM Location&	Article No 25,39&	Neural Network, local learning, Support vector machine, autoregressive models, Particle Swarm& No. 32\cite{li2009sites}&	Particle Swarm Optimization&	First, explain the solution of the problem based on the improved PSO algorithm. There is a fitness value for each place in the map. /the place with the shortest distance from the place with the best fitness is the best place in collection of candidates, which is result we want.\\
\hline
3&	Fraud Detection&	Article No 30,34,36,38,39&	Threshold-based sequence time delay embedding (t-stide), hidden Markov model (HMM), k-nearest neighbor (k-NN), self-learning detection method, Empirical research, Expert System&	No. 30 \cite{anderka2014automatic}&	Threshold-based sequence time delay embedding (t-stide)&	0.85
precision / recall\\
\hline
4&	User Interface&	Article No 3,24&	Process Mining, Pattern Recognition&	No. 24
\cite{omri}&	Pattern Recognition&	Finally it is seen through this paper that the incorporation of biometric features will be essential to ensure that these systems are secure enough.\\
\hline
5&	Customer Behavior&	Article No 4,21,28,33,37&	Multiple logistic regression analysis, Pearson correlation, Genetic Algorithm &	No. 28 \cite{kumbhar2011factors} &	Correlation and Regression analysis  &	cost effectiveness, easy to use and securityand responsiveness were influence customer satisfaction at 36\% variance.\\
\hline
6&	Replenished Strategy&	Article No 9,13,22,23,41,42&	Nearest Neighborhood, Genetic Algorithm, Mix-integer programming model, flexible clustering heuristic&	No. 9  \cite{ekinci2015optimization}&	Nearest Neighborhood&	20\% 
MAPE\\
\hline
7&	ATM Failure&	Article No 10,26,31&	Autoregressive Moving Average (ARMA), Classification &	No. 31 (Zhao, Xu, and Liu, 2007) & a novel methodology to use auto-regressive moving average (ARMA) model &	2.48\% Mean Absolute Error (MAE) \\
\hline
8&	Peak Time&	Article No 12,35&	Adaptive Bayes Network, Support Vector Machine, Native Bayes&	No. 35
\cite{rodrigues2010economic}&	Regression & In general, the results indicate that cash withdrawals are negatively affected the day before the holiday. This effect is even more pronounced during the holiday. Concerning the day after the holiday, the results are less general, pointing on average to a positive effect.\\
\hline
\end{tabular}}
\end{center}
\label{tab:2}
\end{table}

\end{document}