Remove some of the guesswork...

...from travel or even locally.

If you're doing well, stop reading. If you could use a little help or are just curious, keep reading.

In a nutshell, I want to help you predict success. If you hate going to a city and losing money and/or wasting time, I might be able to help. Feel free to keep using intuition and other tools you already have. I would supplement those tools, not replace them. I will analyze data that you provide and provide you with a formula and how to use it. If the data do not provide a useful formula (it happens), I won't charge you. If you have enough data, I can provide a general formula, and I can provide separate formulas by city.

In terms of what I will charge, I'll do a few free ones for early adopters. You can help me in terms of getting the word out and/or helping me set a good price range.

In terms of discretion, I'll send more detail about sending data, but I DON'T need to know how you define success (don't tell me), and the predictors you choose do NOT need meaningful labels (P1, P2, P3, etc.). I understand discretion is important.

Who TF am I? Yeah, I know, joined in 2009, no posts until now. I've been on vacation from the hobby. I still am (so, good news: I won't be asking for freebies or service exchange). But I need extra cash, and I gots skillz (well data analysis skills). Some of the Corpus folks can tell you who I am. Some of you may remember me from other boards.

So, PM me if you want to give it a try.

CCH

Here's a sample (I just made up the numbers) for those who might be confuserated...

City,P1,P2,Success
1,5,1,100
1,4,2,90
1,3,1,80
1,2,2,70
1,1,1,60
1,5,2,100
1,4,1,90
1,3,2,80
1,2,1,60
1,1,2,70
2,5,1,100
2,4,2,80
2,2,1,90
2,3,2,70
2,1,1,60
2,5,2,90
2,4,1,100
2,2,2,70
2,3,1,80
2,1,2,60

Again, I don't need or want to know how you define success. As long as a 0-100 scale is used, I'm good. I also don't need to know what the predictors are, and I don't need to know what the cities are.

Anyway, with the data above, you get a good formula for each city and overall. City 1 data produces the best formula (less error); City 2 data produced the formula with the most error, but again, the data produced three good formulas. NOTE, P2 is not part of any of the formulas -- not a useful predictor.


Overall formula: S = 8.6*P1 + 54.34
City 1 formula: S = 9.5*P1 + 51.5
City 2 formula: S = 7*P1 + 60

How do you use the formula?
For the city one formula and P1 = 5, plug this into google... (9.5*5)+51.5
For the city one formula and P1 = 4, plug this into google... (9.5*4)+51.5
And so on.

So, whatever P1 is, the forumla predicts: the lower P1 gets, the worse you will do in terms of Success. If you are making a decision, and you're not sure, the formula can help you decide.

For those who are still awake, I can provide more detail if you are curious.

CCH

Interesting.. This isn't too be harsh, but this is inaccurate. I applaud what you are trying to do, however. In fact, I kind of like it. Going through introductory statistics many years ago, I was deeply impressed with the multiple regression formula also.

Here is what you need to correct:

1st. You've given out the standard regression formula
2nd. Choosing the right predictors is important. I would never leave this to a client and just plug the formula into google analytics. Two things that are often not taught in introductory statistics are the concepts of multicollinearity and Matrix ill-conditioning.
3rd. You've made an assumption that there will be sufficient data recorded and enough of it to actually meet the prerequisites of having a good data set. In the real world, this is often not the case. The formula doesn't work with a set of data that does not meet requirements.

I'm going to leave the rest of this alone, but I would suggest using an Excel What-IF analysis..

http://www.gcflearnfree.org/excel2010/21

Also, understand that it is very unlikely that the predictions will be accurate with the amount of data that each provider has.

I still applaud you, however, for trying to bring a statistical slant to this business model. I'm not saying this is good or bad; it is just that the possibility of error is very high.

Wow good point

Interesting.. This isn't too be harsh, but this is inaccurate. I applaud what you are trying to do, however. In fact, I kind of like it. Going through introductory statistics many years ago, I was deeply impressed with the multiple regression formula also.

Here is what you need to correct:

1st. You've given out the standard regression formula
2nd. Choosing the right predictors is important. I would never leave this to a client and just plug the formula into google analytics. Two things that are often not taught in introductory statistics are the concepts of multicollinearity and Matrix ill-conditioning.
3rd. You've made an assumption that there will be sufficient data recorded and enough of it to actually meet the prerequisites of having a good data set. In the real world, this is often not the case. The formula doesn't work with a set of data that does not meet requirements.

I'm going to leave the rest of this alone, but I would suggest using an Excel What-IF analysis..

http://www.gcflearnfree.org/excel2010/21

Also, understand that it is very unlikely that the predictions will be accurate with the amount of data that each provider has.

I still applaud you, however, for trying to bring a statistical slant to this business model. I'm not saying this is good or bad; it is just that the possibility of error is very high. Originally Posted by Soonerman12

I understand the assumptions of multiple regression. The example was meant to be simple, and a discussion of statistical assumptions may have detracted from the simplicity.

Starting with the provider's opinion, I'd like to let the data determine which predictors are good or bad. I also understand the data may not yield useful results. I hinted at this in my initial post. However, we won't know unless we try.

Although Excel is not my favorite, I'll take a look (thanks). Sharing the results discreetly and securely, is a concern of mine.