Business Marketing Management

Running head: BUSINESS ADMINISTRATION BUSINESS ADMINISTRATION 1 BUSINESS ADMINISTRATION Table of Contents Introduction ................................ ................................ ................................ ................................ ..... 2 Time - series forecasting models ................................ ................................ ................................ .. 2 Movi ng Average ................................ ................................ ................................ ......................... 2 Exponential smoothing ................................ ................................ ................................ ............... 3 4 quarterly moving average ................................ ................................ ................................ ........ 3 Exponential smoothing ................................ ................................ ................................ ............... 4 Comparison between the 4 quarterly moving average methods and exponential smoothing method ................................ ................................ ................................ ................................ ......... 5 Text Mining in businesses ................................ ................................ ................................ .......... 8 Cluster analysis in business ................................ ................................ ................................ ........ 9 Conclusion ................................ ................................ ................................ ................................ .... 11 References ................................ ................................ ................................ ................................ ..... 12 2 BUSINESS ADMINISTRATION Introduction Time - series forecasting models In this report, it has been already mentioned that two types of time series forecasting methods have been investigated and applied for the given sample data set, Rynair passenger. In this data set, the total number of Rynair passengers has been given every quarter for the given period 2014 to 2019. By using the processes of the time series forecasting methods such as moving average and exponential smoothing the forecasted numbers of passengers for the given period would be estimated. Before the process of e stimations in these two selected forecasting methods, the paper would like to discuss the processes of Moving average and Exponential smoothing briefly. In the second section of this report, the paper would discuss the two different types of the data mining process such as Text mining and cluster analysis. The methods are critically analyzed in the following section of this paper. Moving Average In statistics, moving the average is the most popular method of forecasting. The moving average process of forecasting is mainly used for the time series data (Marshall, Nguyen & Visaltanachoti, 2017) . By this forecasting process, the short - term data will be relatively smoother and it can also able to address the long - term trend (Zhang, Lin & Yang, 2020) . Moving Average has been used in different subjective perspectives as this method has been used in Finance in order to forecast the stocks and bond prices in the long term, in o rder to address return from any investment ventures. The moving average process has been also popularly used on economic grounds. In economics, to estimate the GDP or employment trend, these methods have been used. There are mainly four different forms of moving average forecasting processes as Simple 3 BUSINESS ADMINISTRATION moving average, weighted moving average, cumulative moving average, and exponential moving average. Exponential smoothing Exponential smoothing is another type of statistical technique for forecasting the tim e series data set (HANSUN, 2016) . It is basically used for the univariate data set. By this exponential smoothing method, the forecasted values have been calculated on the basis of lagged forecasted as well as lagged actual values n, 2016) . There are various types of exponential smoothing such as single exponential smoothing, double exponential smoothing, and triple exponential smoothing (Mahajan, Chen & Tsai, 2018) . In order to estimate forecasted values in the exponential smoothin g process, the following formula has been used. Ft = At - 1 + (1 - ) Ft - 1 (Smyl, 2020) Where F represents the forecasted values in the year t and (t - 1) whereas is the given smoothing factor of this forecasting method, here the value of is fixed here. At - 1 shows the actual value for a particular variable in the (t - 1) period. In the following section of this paper, these two - time series forecasting methods have been used for the given data series. It is already mentioned that the given data series, Rynai r passenger contains the data for the 5 years from 2014 to 2015. However, in this paper forecasted values have been only calculated for the last year 2019. The results of forecasts in two different methods have been mentioned in the following. 4 quarterly moving average The selected data set contains quarterly data for the selected time span thus the moving average of the passengers has been calculated on the basis of the quarterly moving average estimation. In 4 BUSINESS ADMINISTRATION this analysis, a 4 quarterly moving average technique has been conducted, and by this process the forecasted numbers of Rynair passengers have been estimated in EXCEL. In this case a formula has been used for calculating the 4 quarterly moving average, i.e. MVt = At - 3 + At - 2 + At - 1 + At/4 (HANSUN, 2 016) . Here, A refers to the actual value for the respective periods. The result is presented in the following Table 1 Moving Average The above table shows the actual numbers of passengers and the forecasted numbers of passengers by applying the 4 quarterly moving averages. Exponential smoothing The number of forecasted Rynair passengers has been estimated here by using the exponential smoothing method. The formula which has been used in order to calculate the forecasted values within this process is in the following Ft = At - 1 + (1 - ) Ft - 1 Where F represents the forecasted values in the year t and (t - 1) whereas is the given smoothing factor of this forecasting method, here the value of is fixed here. At - 1 shows the actual value for a particular variable in the (t - 1) period. Table 2 Exponential Smoothing 5 BUSINESS ADMINISTRATION The above table shows the actual numbers of passengers and the forecasted numbers of passengers by applying exponential smoothing. Comparison between the 4 quarterly moving average methods and exponential smoothing method Now the paper would like to understand that which of the selected time - series forecasting estimate would be relatively more efficient. That will be compared on the method of mean squared error or MSE (Ohno et al., 2017) . By the process of MSE, the deviation between actual and estimated or forecasted values have been addressed. This method mainly reflected the average difference betwe en actual and forecasted squared values (Rougier, 2016) . It is mainly the risk or error of estimation. In order to calculated MSE, the following formula is used. MSE = 1/n ( Gu et al., 2016) Where, xi is the actual and xhat are the predicted values of a sample, and n defi ned the total number of data points. By using this formula, it is calculated that the mean squared error for the 4 quarterly moving average method is 23.13 and the MSE for the exponential smoothing is 41.96. If the value of MSE is relatively lower then the process of estimation is relatively better. From, the above result, it has been observed that the value of m ean squared error is relatively low for the moving average technique thus it is a more appropriate method in this given data set. The following graphs can be described the facts more precisely. 6 BUSINESS ADMINISTRATION Figure 1 Moving Average In figure 1, the red forecasted line is relatively smoother than the blue line which represents the actual number of passengers. Thus, the moving average forecasting method makes the data series relatively smooth. The deviation between the blue and red lines refers to an error in estimat ion. Figure 2 Exponential Smoothing 7 BUSINESS ADMINISTRATION In figure 2, the red forecasted line is relatively smoother than the blue line which represents the actual number of passengers. Thus, the exponential smoothing forecasting method makes the data series relatively le ss volatile. The deviation between the blue and red lines refers to an error in estimation. By comparing figure 1 and 2, it can be observed that the deviation or gap between actual and estimated numbers of passengers lines have been low in the moving aver age method. Thus, the moving average method in the time series forecasting method has been able to predict the number of Ryanair passengers more appropriately. 8 BUSINESS ADMINISTRATION Text Mining in businesses The process of text mining is also popularly recognized as te xt data mining. An organization has been making text mining in order to classify or synchronize its raw data (Roelands, va n Delden & Windmeijer, 2018 ) . By the process of data mining, an organization can able to transform its unstructured raw data into a st ructured meaningful format. In order to do that, the organization has been taken the help of various tools such as Bayes, Naïve, SVM, or Support Vector machine and different deep algorithm methods. After the process of text mining, the organization can dra w several new insights or meaningful patterns from the data that help the organization in order to make any decisions regarding the firm ( Salloum et al., 2017) . Text is the most popular form of database, this type of database can be characterized by three different structures such as structured data, semi - structured data, and unstructured data. Structured data The structured data can be formatted into tabular form, that form consists of many rows and columns. With these numerous rows and columns, data storing would be easier. For this specific structure of data, different data analysis processes and machine learning algorithms are very easily applicable. Examples of this type of structured data are the names of the consumers, their phone numbers, and th eir addresses. Semi - structured data The semi - structured data set is a basic combination or blend between structured and unstructured data. It does not have the capacity in order to mee t all the requirements for the relational relationship of the organiza tions. The common example of this type of database is XML, HTML, and JSON files. 9 BUSINESS ADMINISTRATION Unstructured data This type of database does not have any specific pattern or any predetermined format. This unstructured database includes text sources such as various revi ews of a product, like for the product in a social media or other types of media such as video or audio, etc. (Eberendu, 2016) . In the present day, 80% of the world database has resided to unstructured data. By the various tools of data mining and natural language processing (NLP), the unstructured databases allow modifying a structured database format. Disadvantages However, the process of data mining is associated with various cons in order to structure a database. Web mining is criticized for privacy in vasion. In this process, the privacy of the consumers would lose in order to analyze the data. Another most important concern of this process is that here customers have been de - individualized by addressing their mouse clicks. Cluster analysis in business In cluster data mining analysis, the data sets of an organization have been divided into different parts on the basis of the similarities between the data set. A label has been assigned to the cluster groups which will be helpful to understand the change s by the previous classifications. The objective of the cluster data mining analysis is to address groups of objects which are likely to each other within the cluster whereas the objectives of the groups differ between the various groups ( Fourati - Jamoussi & Niamba, 2016) . There are various uses of data cluster analysis in the field of data analysis, image processing, reorganization of data patterns and market research, etc. The cluster data mining process can be used in the different fields of interest. For instance, th is analysis can be used for the classification of different plants or animals, or in order to identify the 10 BUSINESS ADMINISTRATION specific type of land areas, etc. The cluster of data mining can also use in the field for customers patterns identification. Suppose, a data set contains credit market frauds in a particular area of the banking sector. By applying the cluster analysis of data mining, the pattern of deception can be identified and classified. In the following ways, the process of cluster data mining can se rve different purposes in order to structure a database. By using cluster analysis, the database would be relatively more understandable and easy to interpret ( G udanowska, Kononiuk & D 2020) . As is already mentioned that most of the databases re side in an unstructured form and a messed up format. By this process of grouping, the database is organized into some groups of similar objects. This process is also useful for high dimensional data set as well as small - sized data set. The algorithms of clustering can be used for categorical data, binary data set, and also for interval classified data set. There are some popular methods of data clustering such as Partitioning Clustering Method , Hierarchical Clustering Methods , Density - Based Clustering Method , Grid - Based Clustering Method , Model - Based Clustering Methods , and Constraint - Based Clustering Method . Disadvantages The most important disadvantage of cluster analysis in data mining is that this process is inapplicable for the data set which is h omogenous in nature. On the other hand, the data clustering or the data grouping has also been problematic for a very large - sized data set, for this case, the concept of data clustering would become useless. There are also deficiencies in the 11 BUSINESS ADMINISTRATION existing proc ess of an algorithm for clustering. On the contrary, in this case, all algorithms have been tried to find out some type of clusters and it will fail if the data set contains a mixture of different clusters. Conclusion This report concludes that within the two different time series forecasting methods such as moving average and exponential smoothing, the method moving average can forecast the quarterly number of passengers for the selected sample because the value of MSE is comparatively low for the mov ing average process. On the contrary, in the second section of this paper different processes of data mining such as Text mining and Cluster analysis have been discussed. By the various processes of text mining, a database has been turned to structured or semi - structured data based. On the other hand, by clustering the data mining process, the data set is classified into various groups with the same objects. 12 BUSINESS ADMINISTRATION References Eberendu, A. C. (2016). Unstructured Data: an overview of the data of Big Data. Inte rnational Journal of Computer Trends and Technology , 38 (1), 46 - 50. Fourati - Jamoussi, F., & Niamba, C. N. (2016). An evaluation of business intelligence tools: a Journal of Intelligence Studies in Business , 6 (1). Gu, K., Wang, S., Zhai, G., Ma, S., Yang, X., & Zhang, W. (2016). Content - weighted mean - squared error for quality assessment of compressed images. Signal, Image and Video Processing , 10 (5), 803 - 810. Application of Cluster Analysis for the Selection of Key Competences of Future - Oriented Entrepreneurs. Engineering Economics , 31 (5), 565 - 574. series analysis. Balkan J ournal of Electrical and Computer Engineering , 4 (2), 75 - 78. series analysis. Balkan Journal of Electrical and Computer Engineering , 4 (2), 75 - 78. Mahajan, S., Chen, L. J., & Tsai, T. C. (2018). Short - term PM2. 5 forecasting using exponential smoothing method: A comparative analysis. Sensors , 18 (10), 3223. Marshall, B. R., Nguyen, N. H., & Visaltanachoti, N. (2017). Time series momentum and moving average trading rules. Q uantitative Finance , 17 (3), 405 - 421. 13 BUSINESS ADMINISTRATION Ohno, S., Shiraki, T., Tariq, M. R., & Nagahara, M. (2017). Mean squared error analysis of quantizers with error feedback. IEEE Transactions on Signal Processing , 65 (22), 5970 - 5981. Roelands, M., van Delden, A., & Windm eijer, D. (2018). Classifying businesses by economic activity using web - based text mining . Statistics Netherlands. Rougier, J. (2016). Ensemble averaging and mean squared error. Journal of Climate , 29 (24), 8865 - 8870. Salloum, S. A., Al - Emran, M., Monem, A. A., & Shaalan, K. (2017). A survey of text mining in social media: facebook and twitter perspectives. Adv. Sci. Technol. Eng. Syst. J , 2 (1), 127 - 133. Siami - Namini, S., & Namin, A. S. (2018). Forecasting economics and financial time series: ARIMA vs. LSTM. arXiv preprint arXiv:1803.06386 . Smyl, S. (2020). A hybrid method of exponential smoothing and recurrent neural networks for time series forecasting. International Journal of Forecasting , 36 (1), 75 - 85. nd forecasting with four - parameter exponential smoothing. International Journal of Production Economics , 181 , 162 - 173. Zhang, N., Lin, A., & Yang, P. (2020). Detrended moving average partial cross - correlation analysis on financial time series. Physica A: s tatistical mechanics and its applications , 542 , 122960.