Minimize/Maximize
Entry Details
 Points:7
 Votes:7
 Rank:44
 Unique Views:53
 Total Views:754
 Comments:2
 Favorited:0
Linked Submissions:
Share this entry
      
  •   
  •   
  •   
     
Public voting is now closed -thanks for participating!

Winners will be announced on December 14th, 2015

This Idea is hidden. Change Visibility
Finding the Best Place to Live with SAP Lumira Analytics
(D30438)
Voting is Disabled
Submitted:
Dec 1 2015
Status: 
Winner

WINNER: SAP Partner Medal of Valour



"Best places to live" lists on social media pop up regularly, highlighting areas are ranked based on specific scoring criteria. This approach is flawed, because few of these sites are crunching enough relevant data to provide an meaningful assessment.

 

In this scenario "Best" is a subjective measure that even with well-defined scoring, often leads to poor results.

 

Desirability is a very difficult measure to define and report. In the business world, these problems can exist where definitions of KPIs and the supporting analytics are not well defined, or can be manipulated because there are complexities that are hidden at the surface level.

 

To attempt to analyze the "best" place to live requires 3 realizations:

  • Scoring and ranking geographic areas requires a significant volume of information from different sources, most of which are not downloadable off the web without significant processing and cleansing. 
  • Measuring what is good or bad is a personal experience and not a universal measure, due to the high number of variables.
  • Conditions and expectations can change in near real time, so the analysis is more fluent than concrete.

 

APPROACH

 

The data genius challenge afforded me with the opportunity to re-think criteria that most people consider when considering desirability for living. While a country-wide assessment would have been optimal, my scope was reduced to California to meet some of the data constraints on Lumira desktop, and a better opportunity to give my scoring an eye test, based on my familiarity with the state.  For this Data Genius challenge I tried to quantify desirability based on factors that seemed more universal and tried to push the boundaries of the technology.. 

 

I chose 3 criteria to score and analyze my perfect place to live in California. I had a few more criteria which I lacked the time and resources to complete, but will consider for a future version of this app.

 

1. Living Costs and Population Density:

Metropolitan and suburban living provide dramatic differences in cost of living and lifestyle and can impact cost of living. Cost of housing is a simple leading indicator of demand (desirability to live in a specific area), while population density is a great indicator of preference for lifestyle.

 

 

2. Commute Times:

Where you live in relation to your job, is critical for quality of life. Commuting an hour to work is a norm for some professionals and typically expected for some cities up and down the California coast.

 

 

3. Education:

Primary and secondary education are critical for any family, and demographics alone can not highlight where the best schools reside. Instead, I found a public website where I could import schools based on scores, and link that to our demographic data.

 

 

 

 

 

MY FINDINGS:


POPULATION DENSITY IS A TELLING INDICATOR TO POTENIAL LIVING EXPERIENCE: The greatest volume of high end housing (cost of rent or mortgage over $2000/month) is in San Diego and San Francisco. The granularity is not low enough at zip code, because there is a dramatic difference in population density, cost of housing, and invome levels at a zip-code level. To provide more meaningful analysis of housing and population it must occur at the neighboorhood or block level which is data we have access to, but did not utilize in Lumira Desktop.

 


AVOID LOS ANGELES UNLESS YOU ENJOY DRIVING: Of the 2.38 million people who live in Los Angeles, 13.6% of Los Angeles commutes for 40minutes or more each day.

 

SAN DIEGO BOASTS THE BEST WEATHER AND EXPENSIVE PROPERTY BUT NOT THE BEST SCHOOLS: Of the top 20 schools in California, only 1 (ranked 19) exists in San Diego.

 

SO WHAT IS THE BEST AREA TO LIVE IN CALIFORNIA? Based on my scoring and criteria where cost of housing was high, weight of education quality is low, and after exploration I can live close to the coast with fairly low population density, it was these zipcodes that ranked the highest.. However, my analysis is inconclusive, because I need to feed the machine with more data.

 

PROCESS

These are 5 different and important phases of analysis with Lumira that can require different skillsets and roles to appraoch more complex problems.


DATA ACQUISITION

Knowing what data to obtain and where to find it is half the battle. Having access to premium demographics data and boundaries from CMaps Analytics third party partnerships streamlined the research and data wrangling typically required from public data sources. 


DATA PREP:

To create this analysis there was a mixture of data from different public and premium data sources to paint a realistic picture. The most important facet of data prep is calculating new measures and linking multiple data sources.

 

EXPLORATION:

The most important process of transforming data into Information required interactive exploration where visualizing, ranking and filtering data uncovered interesting trends, and deficiencies in my data. A variety of multi-dimensional visualization options like Treemaps were useful, along with advanced geographic visualizations and imagery from CMaps Analytics were helpful to provide context not available out of the box.

 

STORYBOARD

Communication is one of the most challenging facets for a data wrangler/ analyst. Finding a happy medium between a "storyboard" and a dashboard is quite tricky, when insights are subjective to the end user and the span of possible results is expansive. The ability to quickly pivot and create storyboards is crucial to pain a picture as the performance definition changes.

 

ACTION

A storyboard or dashboard without action is a solution destined for failure. For this particular solution, a key criteria was to drive further research, in this case for more detailed housing analysis outside of SAP Lumira. For this scenario we have concatenated a custom dimension which includes a URL to Zillow for a given zip code for real-time housing availability and pricing.

 

 

 

LOCATION, LOCATION, LOCATION... LESSONS LEARNED:

 

WHERE IS AN IMPORTANT QUESTION

In this scenario "Where" was a primary driver for selecting the best place to live. Just looking at data on a map is not enough, specifically for data exploration. Selecting, filtering, excluding, and searching were critical tools for success to filter out the noise. Additionally, having access to CMaps geocoder allowed me to use City/State and addresses with confidence.

 

 

PRICE SENSITIVITY AS A VARIABLE

Metropolitan and suburban living provide dramatic differences in cost of living and lifestyle. This is one measure that "best places" to live surveys get wrong, and highlight why "best" is a subjective measure.  Cost of living scale is very personal, so to provide an accurate assessment, that variable needs to be exposed to the user via data exploration or built with a  For this analysis we assessed the combination of measures including housing cost and density. Cost is a critical variable so to present a story, we need to indicate 3 different price levels..

MY FINDINGS:


POPULATION DENSITY IS A SIMPLE LEADING INDICATOR TO THE DESIRED LIVING EXPERIENCE BUT NOT ENOUGH: The most expensive areas in California offer a dramatically different experience only visible with street view. 


AVOID LOS ANGELES UNLESS YOU ENJOY DRIVING: Of the 2.38 million people who live in Los Angeles, 13.6% of Los Angeles commutes for 40minutes or more each day.

 

SAN DIEGO BOASTS THE BEST WEATHER AND EXPENSIVE PROPERTY BUT NOT THE BEST SCHOOLS: Of the top 20 schools in California, only 1 (ranked 19) exists in San Diego.

 

SO WHAT IS THE BEST AREA TO LIVE IN CALIFORNIA? Based on my scoring and criteria where cost of housing was high, weight of education quality is low, and after exploration I can live close to the coast with fairly low population density, it was these zip codes that ranked the highest.. However, my analysis is inconclusive, because I need to feed the machine with more data.

Promotes

Demotes

No evaluations ready
Recorded Outcomes
Record an Outcome