valinvescap
New Member
- Joined
- Mar 11, 2014
- Messages
- 2
First, this is long, but I want to be as descriptive as possible. This modeling is explores using basic attendance data to enrich decisions (which folks to call to keep attendance up).
THE PROBLEM/QUESTION: I need help in #excel.I'm using MS Office 2013 and want to model attendance records for a volunteer group to see if any relationships exist between attendance, weather, age, sub-committee assignment, time of meeting, type of meeting, distance from meeting location, tenure (how long person has been in the group), etc. The goal is to first visually present the data so we see if any one group or factor (weather, distance) is the cause for regular absences. Then, going a step farther after the data is conditioned and ready for modeling, use this data set to generate a probability of absence under certain conditions (i.e. forecasting). If the probability of absence breaks a certain threshold, the next step is to reference a list of phone numbers tied to the person, and generate a "call and encourage list".
DATA DESCRIPTION: Right now I have a data conditioning problem. The role came in a typical attendance format: column A had each person's name (A1= student 1, A2 = student 2), dates start at b4 (under header rows described below) and continue across. Also, the time and type of the meetings change, but it came to me in a stacked column format. For example, b1 has a date, b2 has a time, and b3 has a type. So, b1, b2, and b3 are header rows. Some columns include other variables like sub-committee assignment, birthdate, etc.
When the secretaries delivered the data to me, I knew conditioning was needed first. My first thought was turn this into a matrix: make each day an observation where the person gets a "yes" or "no" for attendance, and other characteristics become columns or variables. If I have a group of 130+ folks, and cover 50+ days, that means I'll end up with 5000+ observations.
KEY CONDITIONING QUESTION: How would I covert the data as its described above into what I want (a matrix with observations as rows and columns as variables). How do I do this in minimal time. Right now, I created a basic macro, and go line by line. Any other suggestions?
In this format, I can use Pivot tables. I know how to create a moving window average, set upper and lower limits, and alert when the next day's value goes outside of either the lower or upper threshold. How would I forecast? Exponential smoothing?
The forecasting is where I would like to make a probability estimate of attendance based on past performance. The goal is to create an "alert", so that when conditions for an individual suggest they may be absent (weather, time of meeting, type of meeting, etc.)
-That means I need a web service attached to the sheet for weather. I have addresses for all the people.
-I also need a web-service that automatically calculates the distance between a home address and a common destination (what if my destination is between two places)
-Is there a way for me to create a macro that will read out new information on attendance, and automatically generate an "observation" row in the format I set, so I can continuously model the data?
Any suggestions on how to add the web service for weather, condition the data (attendance was a check mark..I'm adding dates), and/or create a code to take an address from a cell and calculate travel distance to a common destination?
I appreciate any help. If this is tooo hard, I appreciate that as well. It's a volunteer group
THE PROBLEM/QUESTION: I need help in #excel.I'm using MS Office 2013 and want to model attendance records for a volunteer group to see if any relationships exist between attendance, weather, age, sub-committee assignment, time of meeting, type of meeting, distance from meeting location, tenure (how long person has been in the group), etc. The goal is to first visually present the data so we see if any one group or factor (weather, distance) is the cause for regular absences. Then, going a step farther after the data is conditioned and ready for modeling, use this data set to generate a probability of absence under certain conditions (i.e. forecasting). If the probability of absence breaks a certain threshold, the next step is to reference a list of phone numbers tied to the person, and generate a "call and encourage list".
DATA DESCRIPTION: Right now I have a data conditioning problem. The role came in a typical attendance format: column A had each person's name (A1= student 1, A2 = student 2), dates start at b4 (under header rows described below) and continue across. Also, the time and type of the meetings change, but it came to me in a stacked column format. For example, b1 has a date, b2 has a time, and b3 has a type. So, b1, b2, and b3 are header rows. Some columns include other variables like sub-committee assignment, birthdate, etc.
When the secretaries delivered the data to me, I knew conditioning was needed first. My first thought was turn this into a matrix: make each day an observation where the person gets a "yes" or "no" for attendance, and other characteristics become columns or variables. If I have a group of 130+ folks, and cover 50+ days, that means I'll end up with 5000+ observations.
KEY CONDITIONING QUESTION: How would I covert the data as its described above into what I want (a matrix with observations as rows and columns as variables). How do I do this in minimal time. Right now, I created a basic macro, and go line by line. Any other suggestions?
In this format, I can use Pivot tables. I know how to create a moving window average, set upper and lower limits, and alert when the next day's value goes outside of either the lower or upper threshold. How would I forecast? Exponential smoothing?
The forecasting is where I would like to make a probability estimate of attendance based on past performance. The goal is to create an "alert", so that when conditions for an individual suggest they may be absent (weather, time of meeting, type of meeting, etc.)
-That means I need a web service attached to the sheet for weather. I have addresses for all the people.
-I also need a web-service that automatically calculates the distance between a home address and a common destination (what if my destination is between two places)
-Is there a way for me to create a macro that will read out new information on attendance, and automatically generate an "observation" row in the format I set, so I can continuously model the data?
Any suggestions on how to add the web service for weather, condition the data (attendance was a check mark..I'm adding dates), and/or create a code to take an address from a cell and calculate travel distance to a common destination?
I appreciate any help. If this is tooo hard, I appreciate that as well. It's a volunteer group
