Difference between revisions of "ANLY482 AY2016-17 T2 Group10 Analysis & Findings: Analysis"
Jxsim.2013 (talk | contribs) |
Jxsim.2013 (talk | contribs) |
||
Line 58: | Line 58: | ||
<!-- Body --> | <!-- Body --> | ||
− | ==<div style="background: #ffffff; padding: 17px;padding:0.3em; letter-spacing:0.1em; line-height: 0.1em; text-indent: 10px; font-size:17px; text-transform:uppercase; font-weight: light; font-family: "><font color= #000000><strong>Analysis</strong></font></div>== | + | ==<div style="background: #ffffff; padding: 17px;padding:0.3em; letter-spacing:0.1em; line-height: 0.1em; text-indent: 10px; font-size:17px; text-transform:uppercase; font-weight: light; font-family: 'Century Gothic'; border-left:8px solid #1b96fe"><font color= #000000><strong>Analysis</strong></font></div>== |
===<div style="background: #1b96fe;padding:0.6em; letter-spacing:0.1em; line-height: 0.7em; border-radius:20px; font-size:15px; text-transform:uppercase; font-weight: light; font-family: 'Century Gothic'; border-left:8px solid #1b96fe; display: inline-block; margin-bottom:10px"><font color= #fff><strong>BY THERAPY GROUP</strong></font></div>=== | ===<div style="background: #1b96fe;padding:0.6em; letter-spacing:0.1em; line-height: 0.7em; border-radius:20px; font-size:15px; text-transform:uppercase; font-weight: light; font-family: 'Century Gothic'; border-left:8px solid #1b96fe; display: inline-block; margin-bottom:10px"><font color= #fff><strong>BY THERAPY GROUP</strong></font></div>=== | ||
Line 151: | Line 151: | ||
# Polyclinics behave the most oddly, with changing of interactions from medium to high having a significant improvement in mean yet there is no conclusive evidence that a change from low to medium interactions result in a change in mean revenues. | # Polyclinics behave the most oddly, with changing of interactions from medium to high having a significant improvement in mean yet there is no conclusive evidence that a change from low to medium interactions result in a change in mean revenues. | ||
</div> | </div> | ||
+ | <!-- End Body ---> | ||
− | <!-- | + | <!-- Body --> |
+ | ==<div style="background: #ffffff; padding: 17px;padding:0.3em; letter-spacing:0.1em; line-height: 0.1em; text-indent: 10px; font-size:17px; text-transform:uppercase; font-weight: light; font-family: "><font color= #000000><strong>Further Analysis, Quarter Anomalies</strong></font></div>== | ||
+ | <div style="margin:0px; padding: 10px; background: #f2f4f4; font-family: Century Gothic, Open Sans, Arial, sans-serif; border-radius: 7px; text-align:left; font-size: 15px"> | ||
+ | For each | ||
+ | therapy group | ||
+ | , the team has decided to explore deeper by looking at each therapy | ||
+ | group by quarterly performance | ||
+ | and | ||
+ | each sales channel by a quarterly view. We expect | ||
+ | quarters to follow | ||
+ | the results as in the year’s results, | ||
+ | which otherwise, could mean that | ||
+ | there | ||
+ | is a certain action causing an anomaly during | ||
+ | the quarter. | ||
+ | </div> |
Revision as of 14:09, 15 April 2017
Contents
ACTUAL METHOD: Analysis of Variance (ANOVA) using Fit Y by X
Analysis of Variance is a statistical method used to analyze differences among group means and their variances among and between groups. It is also a form of statistical hypothesis testing to test whether differences between pairs of group means are significant or not.
Prior to using ANOVA, we have attempted using linear regression to generalize the relationship between number of interactions and sales revenue. However, low R-squared values that suggest weak correlation and model not fitting the data were obtained, and these prompted us to carry out similar analysis using nonparametric tests like ANOVA.
The primary step to carry out ANOVA is to discretize our explanatory variable - “interaction count” into bins and as such, converting it from a numerical to categorical variable. The objective of discretization is because we wish to understand whether each of these interaction bins have significant differences between one another when it comes to sales revenue (response). To define the range of interaction counts for “Low”, “Medium” and “High” interaction bins, we consulted our sponsor, who proposed that “Low” is for interaction count less than or equal to 1, “Medium” is for interaction count from 2 to 4 and “High” is for interaction count 5 and above.
Analysis
BY THERAPY GROUP
Therapy Group 1: Adult Vaccines
Results show that the p-value between the different categories is way lower than 0.05, which signifies that a change in interactions from low to high and low to medium results in a change in mean. Between the low to medium category though,the p value is higher than 0.05,signifying there is no conclusive evidence that a change in interaction level from low to medium has any impact on mean revenue levels.
Therapy Group 2: Dermatology
Results show that the p-value between the different categories is way lower than 0.05, which signifies that a change in interactions from low to high and low to medium results in a change in mean. Between the low to medium category though, the p value is higher than 0.05, signifying there is no conclusive evidence that a change in interaction level from medium to low has any impact on mean revenue levels.
Therapy Group 3: Allergy
Results show that the p-value between the different categories is way lower than 0.05, which signifies that a change in interactions from low to high and low to medium results in a change in mean. Between the low to medium category though, the p value is higher than 0.05, signifying there is no conclusive evidence that a change in interaction level from medium to low has any impact on mean revenue levels.
Therapy Group 4: Pediatrics Vaccines
Results show that the p-value between the different categories is way lower than 0.05, which signifies that a change in interactions from low to high and low to medium results in a change in mean. Between the low to medium category though, the p value is higher than 0.05, signifying there is no conclusive evidence that a change in interaction level from medium to low has any impact on mean revenue levels.
Therapy Group 5: Urology
Results show that the p-value between the different categories is way lower than 0.05, which signifies that a change in interactions from low to high and low to medium results in a change in mean. Between the low to medium category though, the p value is higher than 0.05, signifying there is no conclusive evidence that a change in interaction level from medium to low has any impact on mean revenue levels.
Therapy Group 6: Respiratory
Results show that across all categories, there p-value falls in the acceptable range, which means that there is no conclusive evidence that change in interaction levels have an impact on the mean of revenue levels.
Concluding Analysis for ANALYSIS: BY THERAPY GROUP
Across most therapy groups (5/6), apart from respiratory, it seems that the change from low to high interactions and low to medium interactions bears positive and significant impact on the difference in means, which means that it is always better to perform high level interactions for each therapy group than low and medium level interactions.
The change in means reflected is highest on the Urology group. The Respiratory group is an exception, with all p values falling in levels that show that there is no conclusive evidence that a change in interaction can cause any change in mean.
BY SALES CHANNEL
Sales Channel 1: Pharmacies
Results show that across all categories, the p-value falls in the acceptable range, which means that there is no conclusive evidence that change in interaction levels have an impact on the mean of revenue levels.
Sales Channel 2: Private Hospitals
Results show that changing the number of interactions from low to medium has a p value that is more than 0.05 which means that there is no conclusive evidence of a change in means. However, upon switching interactions to high from low, the p value drops to below 0.05, which signifies a conclusive change in mean. There is also no conclusive evidence that a change in means occurs when interactions are switched from medium to high.
Sales Channel 3: Restructured Hospitals
Results show that across all categories, the p-value falls in the acceptable range, which means that there is no conclusive evidence that change in interaction levels have an impact on the mean of revenue levels.
Sales Channel 4: Polyclinics
Results show that changing the number of interactions from medium to low has a p value that is more than 0.05 which means that there is no conclusive evidence of a change in means. That is similar when changing the interaction level from low to high. However, upon switching interactions to high, from medium the p value drops to below 0.05, which signifies a conclusive change in mean.
Sales Channel 5: Specialists (Neighborhood clinics)
Results show that changing the number of interactions from low to medium, low to high and medium to high, all carry p values of less than 0.05, which means that there is conclusive evidence that the means changes across all those categories.
Sales Channel 6: General Practitioner (Neighborhood clinics)
Results show that changing the number of interactions from low to medium and medium to high carry p values of less than 0.05, which means that there is conclusive evidence that the means changes across all those categories. However, a change from low to medium level interactions carry a p-value of more than 0.05, which means that there is no conclusive evidence that there is any change in means.
Concluding Analysis for ANALYSIS: BY SALES CHANNEL
A few findings can be seen from these results:
- Sales interactions increases results in significant improvement in means for neighbourhood clinics, both Specialists and General Practitioners.
- Restructured hospitals and Pharmacies have not been proven to be affected by any change in interaction.
- Different sales channels have extremely different patterns of sales (i.e. affected by interactions in different ways
- Private hospitals only see a change in revenues when interactions are increased greatly from low to high, but not from a low to medium level.
- Polyclinics behave the most oddly, with changing of interactions from medium to high having a significant improvement in mean yet there is no conclusive evidence that a change from low to medium interactions result in a change in mean revenues.
Further Analysis, Quarter Anomalies
For each therapy group , the team has decided to explore deeper by looking at each therapy group by quarterly performance and each sales channel by a quarterly view. We expect quarters to follow the results as in the year’s results, which otherwise, could mean that there is a certain action causing an anomaly during the quarter.