Variables:

Column 1: datestring

Column 2: 0-3_hr_verification

Column 3: 3-6_hr_verification

Column 4: 6-9_hr_verification

Column 5: 9-12_hr_verification

Column 6: 0-6_hr_verification

Column 7: 6-12_hr_verification

Column 8: 0-12_hr_verification

Column 9: normalized_wind

Column 10: tempadv

Column 11: 700_mb_geos._dir

Column 12: 700_mb_geos._mag

Column 13: u_bar_at_den

Column 14: u_bar_at_location_

Column 15: 500-700_mb_geostrophic_shear

Column 16: 700_measured/700_geostrophic_wind

Column 17: 700_measured_-_700_geostrophic_wind

Column 18: relative_humidity

Column 19: cross-mountain_height_diff.

Column 20: static_stability_ratio

Column 21: froude_height

Column 22: scorer_parameter

Column 23: char._imped._ratio

Column 24: lowest_trop

Column 25: local_trop

Column 26: garbage_variable

Column 27: postfrontal_parameter

Column 28: sangster_parameter

Predictors:

Column 10: tempadv

Column 11: 700_mb_geos._dir

Column 12: 700_mb_geos._mag

Column 13: u_bar_at_den

Column 14: u_bar_at_location_

Column 15: 500-700_mb_geostrophic_shear

Column 16: 700_measured/700_geostrophic_wind

Column 17: 700_measured_-_700_geostrophic_wind

Column 18: relative_humidity

Column 19: cross-mountain_height_diff.

Column 20: static_stability_ratio

Column 21: froude_height

Column 22: scorer_parameter

Column 23: char._imped._ratio

Column 24: lowest_trop

Column 25: local_trop

Column 26: garbage_variable

Column 27: postfrontal_parameter

Column 28: sangster_parameter

1) Use an appropriate parametric test to determine if prefrontal and postfrontal windstorms have the same variance in peak wind speed. (α=0.05).

2) Formulate 95% bootstrap confidence intervals on the means of each of the individual predictors (Column 10-28, minus Column 26) for both the prefrontal windstorms and the postfrontal windstorms (separately). Use these results to determine which of the predictors are statistically significantly different between the prefrontal and postfrontal storms. Create a table with all confidence intervals to summarize the results (you can use any software you wish, including Excel, for this table). Briefly discuss the results.

- boot(dataset,statistic,R=1000)

- install.packages(‘boot’)

- library(‘boot’)

- mean.boot <- function(x,d) { return(mean(x[d])) }

3) Use an rbind to join the prefrontal and postfrontal predictor data (Column 10 – Column 28, omit Column 26) into a single matrix (by row). Also, join the prefrontal and postfrontal wind speeds (Column 8) into a single vector (use the c() function), and cbind that with the larger matrix. Your new matrix should have the same number of columns as your beginning dataset, plus 1, and its row dimensionality is equal to the number of cases.

4) Perform two separate stepwise regressions, one on the prefrontal data and one on the postfrontal data. Are the variables you kept the same for both? Which seems to be better based on the summary and ANOVA statistics?

5) Using your larger matrix you created at the beginning, do a classification to determine the following three predictors’ ability to classify prefrontal and postfrontal storms. Use the Sangster parameter (Column 28), the Froude height (Column 21), and the 700mb geostrophic wind speed (Column 16). Assign prefrontal storms as 1’s and postfrontal as 0’s. Use logistic regression for this analysis. Create a contingency table and contingency statistics and interpret the quality of the model.

6) Use the full predictor matrix created in 3 to formulate a hierarchical cluster analysis using Ward’s method. Provide the dendrogram. How many groups do you think there are? Interpret the results.

**Subject Mathematics Statistics-R Programming**