triedandtested
New member
- Joined
- Oct 23, 2012
- Messages
- 1
Hello all. My question is how do I determine the distribution of a given sample? I have an assignment to do and I am stuck....
Scenario: A business has recorded when a customer has arrived (date and time) and has given you the records for a particular day. (The only change here is that the time of arrivals change, the date does not change)
question: Create a Histogram of the Arrival and give a 5% significance test to test whether it is coming from an exponential distribution.
I am stuck here as all I have to use is a bunch of times (date is the exact same for each one), i.e. 12:13am, 12:21am, 12:36am, 12:46am, 1:15am, 1:45am, etc (it goes on for over 100 others). You cannot really make a Histogram with only one variable (time) so I thought maybe time classes???? Like have a class of every 30mins, 2 hours or 3 hours and then show the number of recorded customers arriving and then plot the graph? (i.e. 12:00-12:30am has a frequency of 2 from the example data above).
I believe it is suppose to be a Poisson Distribution, being Lambda = number of events in a given time interval (in this case, within 30mins? or 1 hour? or 2 hours? etc.?) And would I then be able to determine an exponential distribution from the time between each call like:
How does an exponential distribution suppose to look like? Have I done enough to figure out the significance level for the exponential distribution?
How do I find the significance level of an exponential distribution? I guess I can use the Chi Square Test ((observed-expected)^2/expected) but then what is the expected value suppose to be? Is it suppose to be just the average of the time between arrivals? Or does the expected value change because an exponential is suppose to go high on the left side of the graph and then decrease to the right, meaning that the expected value at the start would be higher and gradually decrease?
How would you manipulate the data given to make an exponential distribution?
Really could use some help to put me on track.
Thanks.
Scenario: A business has recorded when a customer has arrived (date and time) and has given you the records for a particular day. (The only change here is that the time of arrivals change, the date does not change)
question: Create a Histogram of the Arrival and give a 5% significance test to test whether it is coming from an exponential distribution.
I am stuck here as all I have to use is a bunch of times (date is the exact same for each one), i.e. 12:13am, 12:21am, 12:36am, 12:46am, 1:15am, 1:45am, etc (it goes on for over 100 others). You cannot really make a Histogram with only one variable (time) so I thought maybe time classes???? Like have a class of every 30mins, 2 hours or 3 hours and then show the number of recorded customers arriving and then plot the graph? (i.e. 12:00-12:30am has a frequency of 2 from the example data above).
I believe it is suppose to be a Poisson Distribution, being Lambda = number of events in a given time interval (in this case, within 30mins? or 1 hour? or 2 hours? etc.?) And would I then be able to determine an exponential distribution from the time between each call like:
- 8 mins between 12:13am and 12:21am, then
- 15mins between 12:21am and 12:36am
How does an exponential distribution suppose to look like? Have I done enough to figure out the significance level for the exponential distribution?
How do I find the significance level of an exponential distribution? I guess I can use the Chi Square Test ((observed-expected)^2/expected) but then what is the expected value suppose to be? Is it suppose to be just the average of the time between arrivals? Or does the expected value change because an exponential is suppose to go high on the left side of the graph and then decrease to the right, meaning that the expected value at the start would be higher and gradually decrease?
How would you manipulate the data given to make an exponential distribution?
Really could use some help to put me on track.
Thanks.