Final Project Presentation

Shazia Khan
May 27, 2015

3 Leadership Styles by Kahan

I like how Kahan compares the leaderhsip styles to traveling:

 http://velvetchainsaw.com/2011/09/14/using-rear-view-mirror-see-ahead/
  • 1. Rear View
  • 2. The Headlights
  • 3. The Trip Advisor

1. Rear View

rearView

Data collected is what happened in the past and Kahan said it is like driving your car while looking in the rear view mirror.

2. The Headlights

headlights

Leaders that rely on what is immediately ahead for their organization, by focusing on customers needs, Kahan says this is like focusing on where your headlights shine. What about beyond the headlights?

3. The Trip Advisor

tripAdvisor

This is looking into the Future. Where do you want to be? What is your destination? It's similar to consulting a travel advisor or map to see where you are going when you take a trip. Travel advisors and futurists have an eye on what's ahead.

Motivation

To use the data to predict the needs of the business and to guide descisions to accompolish goals.

Graph

What are the optimum number of calls that need to be made to get sales, based on the data available.

optimumResults

Methodology

An attempt to find the optimal number of calls to be made on an account in a month by a sales representative.

  • 1) Import data
  • 2) Scrub data
  • 3) Prepare data
  • 4) Explore data
  • 5) Analyze data
  • 6) Draw conclusions

1. Data

The calls and sales data for the past 3 years was acquired from work, in a CSV format. The data is posted on GitHub.

strdata

2. Scrub data/ 3. Prepare data

As the data was acquired in the dataframe the data was explored, scrubbed and prepared for analysis.

  • Added longitude and latitude to plot data on ggmap
  • Merged data and Normatlized data to show calls and Quantity on a row for each hospital
  • Delete duplicate columns

4. Explore data

  • Visual representation of Calls and Units:

    callsggmap unitsggmap

4. Explore data

  • The distribution of Calls and Units:

    CallsDis UntisDist

4. Explore data

  • A lot of the accounts that had orders did not have any calls or had minimum calls which skews the analysis.
  • Correlation when Units > 0 = 18%
  • Correlation when Calls > 0 = 13%

5. Analyze data

  • Q-Q Plots, In general, the basic idea is to compute the theoretically expected value for each data point based on the distribution in question. If the data indeed follow the assumed distribution, then the points on the q-q plot will fall approximately on a straight line.

    qqplotUnits qqplotCalls

5. Analyze data

  • The plot gives an idea of whether there is any curvature in the data.
  • If the red line is strongly curved, a quadratic or other model may be better.
  • In this case, the curvature is not strong but is a straight line, so a quadratic component in the model is not necessary. ResidualvsFitted

5. Analyze data

  • The plot is used to check if the variance is contant.
  • If the red line is strongly tilted up/down, that is a red flag.
  • There are no issues with that in this data - the variance appears constant. scaleLocation

6. Draw conclusion

  • Linear Regression along with ggplot helped to draw a few conclusions:

    sumLM

6. Draw conclusion

ConclusionGraph

  • The graph shows t-value of 20.91 is where the Units are at optimum level.
  • The graph allows me to conclude that for every order we need to make two calls.

Challenges

  • To calculate the optimum number of calls, all the variables that influence the sales order should be identified.
  • Data from marketing department needs to be collected to identify how the promotions affect the sales.
  • It is important to learn predictive modeling to come up with a more satisfactory conclusion.
  • It is important to come up with a number of optimum calls on an account or categorize accounts with optimum calls.
  • Finding data for the need by zipcodes should be acquired to verify that the Product sells even without calls.