Mojahed Ibrahim

The Venn diagram strategy seems to big at Netflix, as it has used the same scheme while deciding the price for Favela Rising as it chose the common customers for both City of God and Born into Brothels which was estimated to be 250000 and made the decision on price. Cinematch was the biggest data analytic application for Netflix. The way it used data from its employees to see how they react if the DVD doesn’t reach on time was interesting.

This might align with today’s class.

http://www.informationweek.com/software/information-management/metlife-uses-nosql-for-customer-service-breakthrough/d/d-id/1109919?

Very creative use of big data. This makes you think of other possibilities and using different data sources and merging them together to make a very meaningful dataset. I think a lot of other industries change along with data analytics, healthcare being a common one. Another one is the finance industry. They use data analytics in different ways to reevaluate positions in the markets, whether in mutual funds or ETFs, hedge funds, or options.

We have seen how sports and entertainment are changing (Chip Kelly article and Netflix discussion and the above link). All industries are changing with the new era. Most easily seen is the sales and marketing departments who now have better information on to whom to market new and existing product lines – and who is buying or making purchasing decisions. Companies are shifting from guesswork when it comes to strategic competitive maneuvers. Data will show precisely how a competitor is aligning in the market. Accounting is even influenced as better data will allow internal accounting departments to gain a better understanding of industry benchmarks (other companies will not give their information up easily). According to Forbes, data analytics is changing the competitive landscape and has become the top priority in aviation, wind, and manufacturing industries.

http://www.forbes.com/sites/louiscolumbus/2014/10/19/84-of-enterprises-see-big-data-analytics-changing-their-industries-competitive-landscapes-in-the-next-year/

What was the total cost in the years 2009 and 2010?
In 2009, the sum of cost is 10338384.53 and in 2010 it is 23696862.69.

What is the most likely time of day for someone to arrive at the emergency room complaining about chest pain?
4:00pm

Total number of people admitted at the hospital for allergic reaction–187 (female-151 and male-36)

Question: During what month do these least number of patients visit the ER? What is the most common visit reason during this month?
Answer: November- difficultly breathing

What is the most common visit reason for patients aged 70 and over?
Answer: Difficulty breathing

Question: Which month had the least amount of patients arriving to the ER by walking? Which month had the highest amount of patients arriving by walking to the ER? How many patients for each month?
Answer: September, 1 patient arrived by walking to the ER; May, 96 patients arrived by walking to the ER.

How many patients visited Emergency Room? Ans: 17580
How many of them were male and how many female? Male : 6867 , Female: 10713
How many of them had Headache? Male : 217 , Female : 521

Q: What reason has the average longest duration of stay for women? How much?
A: Chest Pain, 0.51 days

What is the ratio of 30 year old men to that of women ? Answer: 79:54 (158 men and 108 women who are 30 years old)

Which visit reason has the highest number of high urgency visits?
Chest pain

Q: What are the top three most common reasons for visiting the ER?
A: Chest Pain (2,285), Difficulty Breathing (1,508), and Pain, Abd General (1,352)

What was the average hospital duration of males and females at the age of 53?
Males = 17 minutes
Females = 22 minutes

Q: What is the average stay duration for people aged 29 years old?

Answer: 29 year old stay on average 16.08 minutes, where male patients spend on average 14.4 minutes and female patients spend 17.01 minutes.

Q.How many visits are made by people 20 and younger. What is the most common reason.
A. 651 in total. Flu Symptoms is highest at 85.

Question: What is the average age of the males and females who go to the ER?
Answer: Female: 52.72 years old Male: 52.21 years old.

1) Dart mart example: information regarding no. of employees completed a particular certification.
Parameters: Department, Job titles, and number of attempts

Data mart example: Information on media insertions placed through the agency

Facts: Estimate ID, Property/Station ID, Weeks Active

Measures: Quantity Ordered, Spend

Dimensions: Client, Media Type (Product), Time

Data mart example: Information on student’s contact information and interests
Facts: Address, Phone #, Email, Social media use
Measures: How many students are interested in what topics and where they are located
Dimensions: Group students together by region and interests

For the Data Mart Exercise, the project that I was working on for my internship was looking to set up several data marts for the governmental organization we were consulting for. In terms of facts we were looking at dollar amounts for contracts that were awarded, how much had been set aside for research topics, researcher names, contact information and research subject. Measures were variables such as awards or research citations for research that had award money tied to it, or how much money had been spent. Finally the dimensions were primarily measured in financial terms.

Q: What is the most expensive month on average to go to the emergency room?

A: August 2010 = $1,978.49

What is the total number of males admitted for allergic reaction?

How many teens went to the ER with the highest urgency in 2010?

2

What was the average ER length of stay for 65 year old + male or female? Male, .576 days

While working at Merck this past summer, I dealt extensively with several Merck data marts, including a data mart of the sales of pharmaceutical representatives. Measures included sales in particular regions, number of calls, and doctor visits. Dimensions were primarily financial. This data mart served to evaluate sales representatives, and helped to make sure different reps didn’t cross over each other’s territory.

Data.gov could encourage better use of data by doing the following:
1. Simplicity: Create the website and data interface with advance search features to ensure using the site is intuitive and easy
2. Nomenclature and Classification: Give detailed description of data sets , identify source of data, define meta data to help users understand the data set better
3. Application: For each data category display some useful case studies on the Data.gov website.

If I were in charge of data.gov, I would encourage people to create and share visualizations of the data they cared about most. This would encourage collaboration and make the data sets even more readable by non-technical users. The maximize this potential, rewards and incentives could be given to those with the best/ most accurate/ most useful interpretations of data.

This mashup shows the correlation between high school graduation rate and teen birth rate. Note that those states with the highest ranking (darkest color) are those with either a high percentage of high school graduates, a low teen birth rate, or a combination of the two. The opposite goes for those states with the lowest ranking.

http://www.datamasher.org/mash-ups/people-25-yrs-who-completed-high-school-divided-teen-birth-rate-1000-population

The following link shows the mashup on unemployment and poverty rate in the US. The states with higher resultant figure (product of unemployment and poverty rate) are shown with darker colors in the map and vice-a-versa.
http://www.datamasher.org/mash-ups/unemployment-times-poverty-rate#map-tab

To encourage people to use data.gov, I would make different competitions that would require the use of Data.gov sites and data. We’re a culture that’s all about competition so I think this would be a good motivator for the U.S. I would work with different universities or organizations to see if they could facilitate the competitions and in return get additional federal or state aid. Getting corporate and organizational sponsorship would help with prizes and aid for participants.

The mashup shows the correlation between adult obesity rate and % of adults who are moderately/vigorously active. The darker states have higher correlation between these factors.

http://www.datamasher.org/mash-ups/adult-overweight-obesity-rate-divided-adults-who-were-moderatelyvigorously-active-1

I would encourage healthcare organizations to use Data.gov for insight on trends regarding admission rates, the length of stay and what the return rates are to see what the national average is. They possibly can use the information to make informed decisions concerning their particular organization and see if they are part of that average.

The following graph shows the correlation between the SAT scores and high school graduation rates. The darker states have higher correlation between these 2 factors.
http://www.datamasher.org/mash-ups/sat-scorescombined-reading-mathematics-divided-people-25-yrs-who-completed-high-school

The following mashup shows sat scores times graduation divided by population. The darker states are smarter per capita.

http://www.datamasher.org/mash-ups/sat-scorescombined-reading-mathematics-times-high-school-graduation-rate-divided-populatio-1

This mashup shows the correlation between the percentage of adults who were told they have diabetes and the number of fast food restaurants per
100,000 residents.
http://www.datamasher.org/mash-ups/adults-who-were-told-they-have-diabetes-divided-fast-food-restaurants-100000-residents-0

This may be outlandish, but if I were in charge of data.gov I would hold an open competition to have individuals run the data and find inefficiencies with the government. It seems like something that would probably resonate with a lot of people, and the reward (while also monetary) is to help the government operate better. This in my mind would certainly generate a lot of attention, but potentially cause someone to lose their job!

If I were in charge of data.gov I would advertise the publicity of data being shown around the internet. I would really try and drive traffic towards the data.gov website and encourage people to use this data freely. Once people arrive at the site data.gov should provide tutorials and walk-thrus for people who would like to learn more about how to properly utilize data. I would also encourage people to share their data outputs for some sort of incentive.

If I were in charge of Data.gov, I would hold events akin to Hack-a-thons, where the open source data could be used and analyzed to solve real governmental issues. This would both solve issues and raise awareness of the valuable resource at the same time.

For data.gov I would make it more visible by advertising it so people know it exists. Also possibly taking user feedback to improve it as well.

If you were in charge of data.gov, how would you encourage people to make better use of its data?

Data.gov needs a radical makeover. This should include a rebranding of the website to include an easier process to obtaining information, as well as including a tutorial section to support new users. Not only does the data.gov website need to improve, but so does the marketing strategy to promote the website. Data.gov should take a few notes from the ‘Obama Care’ marketing strategy.

If I were in charge of data.gov, I would create a portal for people to share their creations based on the government data. This would be split into sections based on the focus of the data and how it is used. I would also consider running a contest for data.gov, awarding a prize to the most innovative use of the government’s data. Ideally, this would result in an innovation that could be used to improve government process, and the prize money would be minuscule in comparison to the cost savings resulting from the innovation.

This mashup shows the relationship between the presence of a loaded gun in a household and deaths by firearm within state populations (per 100k people):

http://www.datamasher.org/mash-ups/households-loaded-firearm-times-deaths-due-injury-firearms-100k-0

http://www.datamasher.org/mash-ups/number-births-times-population-covered-health-insurance#table-tab

This is a data mash of number of births times population covered by health insurance. This could help healthcare companies identify target markets based on knowing the largest pockets of the future population that will be covered by health insurance.

So, if I were in charge of data.gov I would take some pretty innovative measures in order to encourage folks to make better use of the data available. I would ensure to hold an annual national data analytics competition and offer monetary compensation. Furthermore, I would have tiered contests for different age groups and have challenges targeted directly to children, thereby fostering their interest in data.

Q1: I think the problem with data.gov is not a lot of people are aware about the existence of it. In order to make better use of data, firstly, they need people to be aware about the existence, for which data. gov should advertise. It can also incentivize people to provide examples for creative usage of the data on its website. In this way, people visiting their website will be aware and motivated to use the data in a better way.

If I were in charge of data.gov, I would initiate an analytics challenge where people are invited to make a case on how to improve governance with regards to delicate issues in their municipality or state. These cases would then be posted in a separate website, and other members of the community can vote on them. The cases would need to rely on really good evidence in order to quality. This way, the governing bodies can have an idea of what sentiments are being shared by their communities and have an opportunity to improve their practices and provide an opportunity for a truer democracy.

1. If I were in charge of Data.gov, I would take the following steps:
a). Agree with Andy’s idea of holding a competition, this opens up the loop holes and helps me improve the website.
b). Classify data according to categories, provide sufficient explanation to headers and he process through which it was collected.
c) Make data easy to analyze (having proper headers in CSV format)
d). Provide basic information about analyzing data.

1)
If I were Data.gov, I would have focused on making people and organisations aware of Data.gov and its usefulness. Having a marketing plan would help Data.gov increase its presence.

The below mashup shows the SAT score compared to the high school graduation rate. States with high high school graduation rate had people with more SAT scores.

http://www.datamasher.org/mash-ups/sat-scorescombined-reading-mathematics-times-high-school-graduation-rate

I have included a mashup of CO2 Emissions by Fossil Fuel Combustion compared with the States population density. What it indicates is that per person, more densely populated states actually produce less CO2 from Fossil Fuel combustion, which was actually the opposite of what I had expected.

http://www.datamasher.org/mash-ups/co2-emissions-fossil-fuel-combustion-metric-tons-co2-divided-us-population-density-state-s-1#map-tab

Yes, you could use this data. Data collected would include resource utilization, productivity, average weekly hours worked, background data, time spent on work activities, time spent on non-work activities, etc. KPIs would include project success rates, time to project implementation, etc.

Yes I believe you can use data from an applicants profile. Our department’s graduate admissions committee looks at specific data in an applicants file to decide to accept them into our PhD program and to offer them an assistantship. The KPI’s would be GRE scores, GPA and strong recommendation letters.

Data from an applicant’s profile can be used to predict their performance in a similar job. The data can be years of experience, skill set, GPA and position. The outcome will be performance rating, productivity, accomplishments, etc.

Q2: IHC case has numerous example which links Data and process. For example at one point the IHC data estimated 15% salary expense was devoted to entering data. The process was then streamlined in 1997 and PCMS processes helped to organize nursing services by task and then structured an electronic data collection system accordingly. PCMS helped reduce redundant data collection.

Q2:
The process is the logical sequence of smaller tasks, the final outcome of a process is the achievement of a result. Each task within a process is performed based on previous data collected and stored in a company’s database. The more accurate and up-to-date the data is, the more accurately it shapes the tasks and the better results are achieved.

Continuing off of our discussion last week on Intermountain Health Care, it is a very distinct relationship that exists between process and data. Intermountain’s original process of gathering and organizing data was unstructured and very time-consuming. They re-vamped the entire data process to reduce redundancy and while cutting back on employees time spent on simple data entry. The process are the steps taken to achieve the overall end goal of data being in a database. By changing the process of data entry, Intermountain Health Care did not change the final outcomes (data), but just the steps taken (or each individual task) in order to get to the final outcomes.

While this is not my job, an example of using data to predict job performance would be the role of help-desk agent. The data in this person’s profile could be level of knowledge in the specific area he is servicing (i.e. degree studied and GPA) and communication skills. The KPIs would be average calls answered per hour, number of times this person got the resolution right the first time, and the satisfaction level of the customer being helped.

Q2:
Yes, data from applicant’s profile could be certainly used to predict performance in a job. Data would be relevant work experience (projects worked on) and educational experience (University attended and highest degree earned). The KPI to measure this would be, the success rate of the projects, job position, and GPA.

You could use data from an applicant’s profile to predict performance. Data such as work experience, length of positions, accomplishments, education experience, and grades, certifications or organizations could all be indicative of performance. KPIs could be cost savings per year, sales per year, and efficiency measures such as process time reductions.

I think data can be used from a persons profile to determine potential job performance. Often times people look for experience at a specific company or industry to understand the applicants skills. For example some employers look for experience at a top consulting firm to indicate a persons ability to approach and diagnose a problem. If an organization can identify what the skills are to do the job well ahead of time it should be easier to recruit for that role. The KPI could be average time for promotion against an organization average or average annual salary growth in comparison to parallel candidates. The KPI should match what the organization is trying to promote so it may very well be “number of product proposals” if the company is seeking to be more innovative, for example.

Q1: I think it is possible to use data, but in many cases the execution is flawed. For example, media buying involves a great deal of math. As a result, we give all new applicants a math competency test to see how they handle “basic math”. The problem is that, in practice, almost all math related to our work is done in Excel or via a third-party software. While it helps to be able to do calculations yourself, there’s enough support that “passing” the math test is actually a poor predictor of performance. I’d like to develop a way to better gauge critical thinking ability as I think my industry is far too conservative and could use a dose of innovation. Unfortunately, I’ve yet to come across any indicator that successfully ties to a candidates competency in this regard.

Q1:

Yes, we could use some data from applicant’s profile to predict the performance. I believe the data could be experience, past projects handled, qualifications, and grades. The KPI (outcome) of this data can be success rate, profitability, and productivity.

Yes I believe it is possible to use applicant data to predict performance. One must be careful, however, because using the wrong data or implementing the wrong process will lead to inaccurate outcomes. One might think that quantitative KPIs are more accurate then qualitative but I would argue that it is situational. As a student applicant coming into school, we are rated on our GPA and GMAT scores. Unfortunately, scores on tests don’t play a direct correlation with final outcomes or deliverables. Qualitative factors also need to be considered heavily. In short, accurate data is important and should be analyzed before using them to predict performance.

Yes, there are some key measures that can be used to predict success in my past role as a business analyst at Merck & Co. Generally, GPA from undergraduate and graduate schools, along with standardized test scores, are a decent approximation of future success. Furthermore, holding certain certifications, such as a PMP, Six Sigma certification, and others can indicate likely success in job functions. However, it is very difficult, because much of job performance can also boil down to commitment from the employee and how much he or she is comfortable in and passionate about their role. The KPI for this scenario would likely be a quarterly peer review of performance.

–I believe you can use data from an applicant’s profile to predict performance if the jobs are similar. The data set must include: employee tenure at previous establishments, sales/revenue performance, project completion timelines, employee hours log sheet, positive vs. negative references, total amount of experience, a coding system for comparable tasks, and interview traits (i.e. smile, intelligent communication skills, body language, appearance). The KPIs would be productivity, profitability, and positive sentiment towards such candidate.

The objective of a process is to collect data and analyze that data to improve the process. The better we are at improving the usability of data, the better our processes will become.

Q2: Based on our discussion of Intermountain Health Care, how would you explain to someone the relationship between process and data?

In the case of Intemountain Health care, the main process under consideration is that of giving the highest quality care to patients. That said, physicians have traditionally used their own experience and expertise to come up with the best treatment possible. They have also used medical literature available and the patient’s records to come up with an individualized treatment plan. However, the data that Intermountain Health Care is collecting and using gets implemented with this critical process and gives physicians a very powerful decision support tool. This way, the process of giving higher quality treatment and care benefits greatly from the data available by proposing routes with higher potential of effectiveness.

That said, the relationship of process and data in this case is two fold: One one end of the spectrum, process helps populate the system with new data whenever a physician treats a patient and enters his/her info in the system. At the other end, data supports the process by giving access to all of the knowledge residing within the organization to physicians and enabling them to make quicker and better decisions.

Q1: There are several additional screenings that people are including with applicant profiles. In my previous profession, the job required a lot of cold calling and personal interviews. In this case, a Myers-Briggs assessment would be helpful as a key indicator of success might be whether a person is an introvert or extrovert.

I would use text analytics to find people that are most relevant. I would use data related to skills and past work experience (qualitative). I would also link LinkedIn profiles to check for endorsements. Number of endorsements are an indicator of people approving someone’s skill. This would be one KPI. I would also check for certifications that anyone is having.

Q2:
Processes are individual tasks. By collecting and analyzing data (and then acting on your findings), these processes can be improved to produce at a more optimum level. A constant monitoring of data can show how process changes are improving or hurting company performance, financially or otherwise.

Q2. With regard to evaluating the outcome of overhauling clinical-care management at IHC, the process might be seen as successful because it was efficient and also provided good clinical outcomes. However, when you analyze the data from financial statements, you can see that there has been a steady operating income and reveals a different and not so positive picture about the outcome of the overhaul.

Q2:

From the perspective of the case, processes are the actions pursued and achieved by an organization to complete an objective. In one case it could be a process for how to load a patients file, for another it could be the process that is used to determine what treatment a patient requires for a heart problem. Overall, processes are repeatable events that, while unique for each case, generally follow a pattern breakdown that allows you to determine what the best outcome will be.

Data is the result of the process and what is generated whenever an action is taken. Thus, proper monitoring of data can better influence the policy that goes into creating processes or allows you to review processes to see what is successful and what isn’t.

You could certainly use data to predict employee success. KPIs such as number of years of work experience, sales records in previous companies, and number of disciplinary actions taken against that person.