| Want to predict data |
Want to predict our sales with the ads expenses on an state-by-state basis for the smartphone we sell |
Want to know the influence rate of the attribute of the research object with the measured data |
|
State,Ads Expenses,Sales
California,100,2100
Texas,130,2500
New York,150,2800
Florida,120,2300
Illinois,110,2300
Pennsylvania,140,2400 |
Sales is shown by the following formula.
Sales = 11.129 * Ads Expenses + 971.429
Pretty good accuracy!
The measured values (x marked) and the predicted values (solid line) are shown by the following graph. |
Number of measurement,Value of attribute A,Value of object
1st,100,2100
2nd,130,2500
3rd,150,2800
4th,120,2300
5th,110,2300
6th,140,2400 |
Value of object is shown by the following formula.
Value of object = 11.129 * Value of attribute A + 971.429
Pretty good accuracy!
The measured values (x marked) and the predicted values (solid line) are shown by the following graph. |
The partial regression coefficient are
11.428571428571429,
971.4285714285713,
Contribution ratio is 0.8163265306122442.
If a contribution rate is more than 0.8, we show 'pretty good' as the accuracy of the prediction. If it is more than 0.5 and less than 0.8, we show 'good'. If it is less than 0.5, you have to analyze data in another way. |
| Want to know how much data I need to prepare |
Want to know the number of people I should pick up to get an accurate result from a questionaire targeting smartphone users in our sales area |
Want to know the number of measurement to reduce a range of error in case I measure a research object |
Yes |
Acceptable Error,Standard Deviation (a range of data to include about 68% of the total centering on a mean)
5,20
(You can skip the first line) |
Please analyze at least 62 numbers of the measured data. |
Acceptable Error,Standard Deviation (a range of data to include about 68% of the total centering on a mean)
5,20
(You can skip the first line) |
Please analyze at least 62 numbers of the measured data. |
This is a sample size to get a 95% confidence interval in case a population is assumed to follow a normal distribution, an error is 5 and standard deviation is 20. |
| Want to estimate a mean |
Want to estimate hours of use for a smartphone from a questionaire targeting smartphone users in our sales area |
Want to estimate a range of the mean of the research object with the measured data |
|
Responder,Hours of use
Andy,1.5
Billy,2.0
Chris,3.0
Duolon,1.8
Eiji,2.2
Foxy,2.1 |
We can estimate the population of Sales Volume about Responder at the following range
1.5690191472127581 to 2.630980852787242
If a range is too big, it is possible for the sample size to be low or for event itself to have a great variability
You can check a sample size at the above 'Want to know how much data I need to prepare'. |
Number of measurement,Measured value
1st,1.5
2nd,2.0
3rd,3.0
4th,1.8
5th,2.2
6th,2.1 |
We can estimate the population of Measured value about Number of measurement at the following range
1.5690191472127581 to 2.630980852787242
If a range is too big, it is possible for the sample size to be low or for event itself to have a great variability
You can check a sample size at the above 'Want to know how much data I need to prepare'. |
This is a 95% confidence interval in case a population is assumed to follow a normal distribution and a population variance is unknown
If the number of data (n) is more than 25 except for the label line, we use a normal distribution table.
If the number of data (n) is less than 25, we use a t-distribution table with a degree of freedom n-1. |
| Want to know how much data I need to prepare (estimation of ratio) |
Want to know the number of people I should pick up to get an accurate result from a questionaire targeting smartphone users in our sales area |
Want to know the number of data to reduce a range of error in case I measure the observed ratio of a research object |
|
Reduced error rate (0-1.0)
0.05
(You can skip the first line) |
Please analyze at least 385 numbers of the measured data. |
Acceptable Error,Standard Deviation (a range of data to include about 68% of the total centering on a mean)
5,20
(You can skip the first line) |
Please analyze at least 62 numbers of the measured data. |
This is a sample size to get a 95% confidence interval in case a population is assumed to follow a normal distribution and an error is 0.05. |
| Want to estimate a ratio |
Want to estimate a ratio of the number of smartphone users that answered that they are satisfied with your service from a questionaire in our sales area |
Want to estimate a range of the ratio of the research object with the observed data |
Yes |
Overall number,A ratio for estimation (0-1.0)
500,0.25 |
We can estimate the population at the following range
0.2120447632071673 to 0.2879552367928327
If a range is too big, it is possible for the sample size to be low or for event itself to have a great variability
You can check a sample size at the above 'Want to know how much data I need to prepare (For estimation of ratio)' |
Overall number,A ratio for estimation (0-1.0)
500,0.25 |
We can estimate the population at the following range
0.2120447632071673 to 0.2879552367928327
If a range is too big, it is possible for the sample size to be low or for event itself to have a great variability
You can check a sample size at the above 'Want to know how much data I need to prepare (For estimation of ratio)' |
This is a 95% confidence interval in case a population is assumed to follow a normal distribution and a population variance is unknown.
|
| Want to know if there is a gap between data and theoretical values |
Want to know if the models smartphone users have are weighted with specific ones from sales volumes in our sales area |
Want to confirm if a research object shows a hypothetical behavior with the measured data |
Yes |
Models,Sales volume,Theoretical values
Model A,125,100
Model B,110,100
Model C,90,100
Model D,100,100
Model E,100,100
Model F,90,100
Model G,85,100 |
This follows a theory!
The measured values (x marked) and the theoretical values (dots) are shown by the following graph. |
Research object,Observed Number,Expected Values
A,125,100
B,110,100
C,90,100
D,100,100
E,100,100
F,90,100
G,85,100 |
This follows a theory!
The measured values (x marked) and the theoretical values (dots) are shown by the following graph. |
A goodness of fit test was done by using a chi-square distribution
The test statistic is 11.5.
If n is the number of data except for the label line, we uses a chi-square distribution table with a degree of freedom n-1 and a significance level 0.05. |
| Want to compare two groups |
Want to confirm that we have a advantage over a competitor in customer satisfaction for our smartphones from its survey |
Want to confirm that my new method is better than a old method |
Yes |
Responder,Our CS,Competitor CS
Andy,91,89
Billy,97,95
Chris,91,92
Duolon,93,90
Eiji,98,96
Foxy,93,92 |
Obviously Our CS is a bigger mean.
Our CS (x marked) and Competitor CS (dots) are shown by the following graph. |
Number of measurement,Measured Values (new method),Measured Values (old method)
1st,90,89
2nd,97,95
3rd,91,92
4th,93,90
5th,98,96
6th,93,92 |
Obviously Measured Values (new method) is a bigger mean.
Measured Values (new method) (x marked) and Measured Values (old method) (dots) are shown by the following graph. |
We did a test of the difference of the paired population means under the null hypothesis that there is no difference between them.
The test statistic is 2.6655699499159153.
If the number of data (n) is more than 25 except for the label line, we use a normal distribution table.
If the number of data (n) is less than 25, we use a t-distribution table with a degree of freedom n-1. Both significance levels are 0.05. |
| Want to know if the data has a outlier |
Want to exclude a outlier that is possible to be measured wrongly in parts test for smartphones |
Want to exclude a outlier that is possible to be measured wrongly in the measured data of the research object |
|
Responder,Hours of use
Andy,1.5
Billy,4.9,unexpected
Chris,2.0
Duolon,2.5
Eiji,3.0
Foxy,2.3 |
Billy,4.9 is a outlier. Exclude it from the data. |
Number of measurement,Measured Values
1st,1.5
2nd,4.9,unexpected
3rd,2.0
4th,2.5
5th,3.0
6th,2.3 |
2nd,4.9 is a outlier. Exclude it from the data. |
Assuming that a population follows a normal distribution, we did a Smirnov-Grubs test.
The test statistic is 1.851421626658838.
If n is the number of data except for the label line, we use a Smirnov-Grubs test table with a significance level 0.05. |
| Want to take the black and white out of data |
Want to exclude inferior parts in a parts test of smartphones |
Want to determine which group a research object belongs to from the measured data |
Yes |
Production Number,Score,Group
01,90,good
02,80,inferior
03,85,good
04,86,inferior
05,82,inferior
06,86,good
07,85,inferior |
The boundary value between good/inferior is 85.21765528644033. You can regard it as 'good' if you get a new bigger Score than this value. |
Object,Measured data,Group
01,90,group A
02,80,group B
03,85,group A
04,86,group B
05,82,group B
06,86,group A
07,85,group B |
The boundary value between group A/group B is 85.21765528644033. You can regard it as 'group A' if you get a new bigger Mewasured data than this value. |
The boundary is a value that univariate Mahalanobis' generalized distance is the same at each group |