You work for the Ministry of Housing. We are facing a potential housing crisis as we believe we are in a Real Estate bubble. We need to identify a few things for our leadership to help them make informed decisions: 1. 2. What is the rate of increase for homes in Boston in the last year, 2 years and 5 years? Identify the main variables that are impacting the price of a home. Use macro-economic trends and housing market information to predict the average home prices in 3 months, 6 months, 1 year and 3 years. Identify the distribution of homes in Boston and their affordability by classifying them into one of three buckets cheap, normal and expensive. This would mean we'd have to make a prediction about the house's classification even though it's not listed on the market. We'd have many data points to help make the prediction like the average sale price in the area, # of bedrooms, square footage, etc. 2. For each of the four questions above, complete the questions in the flow-chart below to determine which analytic method will be applied to each to solve them. Please show your answers to the questions that are assessed in the flowchart and provide justification of your answer. {4 marks) Five Questions for Selecting a Statistical Method 1. How many variables do you have-just one, more How many than one, or way too may? variables do JUST ONE you have? 2. What is your statistical objective-describe, identify/ classify, compare/test, predict, or explain? 3. What scales of measurement do the variables use? WAY MORE Are some variables dependent DESCRIBE 4. Are there dependent and independent variables? TOO THAN and others MANY ONE independent? 5. Are the samples autocorrelated by location (10, 2D, or 3D) or time/sequence (1D)? Conduct a cluster analysis to select representative variables ES Means, standard ranges, quarties Conduct a principal components NO Distribution Elling analysis, factor analysis, CLASSIFY COMPARE PREDICT correspondence analysis, or Are the EXPLAIN multidimensional scaling to dependent variables icons of varlabs reduce the number of variables Istograms, box plat Filtering One-Population Tea NO autocorrelated? [none] [none] probability plots needed to represent most of the Sorting Tests: Cluster analysis variability Anderson Darling Continuous Son Shapiro-Wilk. Lilifor Heats, control charts DESCRIBE CLASSIFY COMPARE PREDICT EXPLAIN YES DESCRIBE CLASSIFY COMPARE PREDICT EXPLAIN Same as for Single Cluster analysis, Same as for single Discrete scale Same as for sing Same as for Same as for variables discriminant variables variables [none] Cluster analysis, Single variables Single variables multidimensional Cross-tabulations analysis Multi-Population Tes fication trec classificalion trees Comelanons scaling principal Correlations ANOVA, ANCOVA, discriminant analysis discriminant analysis Discrete Scales omponents analysis nanparametric beats nan R. Kenda factor analysis, Discrete Scales spearman H. Kendall pendant wiki Continuous scale Tau, gemma correspondence analysis Regression Continuous Scales Continuous Scala egression, canon correlation Pearson Product Pearson Product Moment Correlation Moment Correlation LOCATION DEPENDENT Time or location TIME DEPENDENT dependent? DESCRIBE CLASSIFY COMPARE PREDICT EXPLAIN DESCRIBE CLASSIFY COMPARE PREDICT EXPLAIN Same as for non Trend-surfaces, warishlas with a autocorrelated Interpolation, tren geostatistics Same as for non Same as for non- Same as for none Smoothing methods, Time-series variables with a variables with a surfaces, goostatistic variogramming with a fire-dependent ulocorrelated variables with a time-dependent autocorrelated variables with a time-dependent ARIMA, spectral are egression, ARIMA, variable as a cation-dependan on-depender variable as a and kriging variable as a grouping variable as a grouping variable as a grouping neural networks spectral analysis variable as a "kriging] grouping factor factor grouping lacior factor grouping factor Question Answers and Justification to the Questions in Flow Chart 2 3 4