applied statistics
To deal with the classification of the categories of data:
Bar chart double bar to facilitate comparison:
Pareto chart: Rank the basis of categorical variables frequency of:
Chart PIE : a pie chart for a sample may be distinguished category data
doughnut chart: FIG ring for a plurality of samples can be distinguished category data
Sequence data: up and down by calculating the cumulative percentages accumulated, which can have two different interpretations
FIG line type can be more conveniently placed on a map, to facilitate comparison
Numerical data packet to be processed:
Univariate packet: This variable must be a small number of discrete values, and
Group from the group:
- Isometric packet EG : 10-20 ; 20-30 ; 30-40
- From isopropyl group EG : 10-70 ; 70-80 ; 80-90 ; 90-140 , different from the packet easily cause deviations appreciated, the frequency density may be expressed, understood avoid bias.
Number of samples to give n- , the number of groups obtained by the following formula K , according to the group number K of { class interval = ( maximum value - minimum value) ÷ the number of group } obtained from the group, and finally drawing a histogram, the histogram can be seen by a point even after the line of symmetry into the data.
Common distribution by the histogram obtained are as follows:
legend:
Yellow: the mode
Red: Median
Green: average
Histogram Histogram:
1. Sample amount reflects distribution histogram;
2. The small sample size (less than 75-100 ) there will be an unstable situation
The number of groups may be employed in different points of view is not uniform distribution to determine their suitability to select the group
When the sample number is infinite, the histogram on the line graph into PDF ( factor PMF ): Probability Density function probability density function, the PDF can be deduced CDF: cumulative distribution function distribution function.
Stem and Leaf Stem-and-leaf plot for a small sample size is not a packet of data, it can not be artificially set from the control, usually 10 or 10 minimum number of conventions.
FIG box can be used for many types of comparative samples, as shown below:
But outliers large impact on the box plot, may cause Misleading , use the formula screen all values to elect and remove outliers outlier.
Based on other statistics FIG box: quartiles of which may be replaced by any desired percentile, i.e. the box can be used to determine the confidence interval of FIG.
FIG vertical (drop-line)
Problems between the two variables:
Two-dimensional scatter plot ( 2D scatterplots)
Scattergram overlapping Overlay Scatter:
Scatter Plot Matrix Matrix Scatter
Between three variables: Bubble bubble chart of FIG.
