Marketing Analytics

A. Data Type 

e.g. data definition (Object, characteristics, judge), data relationships (Y, X), Defect, Continuous, Discrete, full population data, sample etc. Data type classification determine what type of tests can be done in the following steps.

按类型来分析字段

B. Stability

e.g. goodness of measure (reliability, variability, validity, consistency), run charts, sample correlation and consistency tests, data trust, consistency of samples, time effects. Stability of data determine the quality of our data, the scope of samples needed, strength of conclusions, and about the existence of common and special cause variation influencing performance.

数据的稳定性 整体的一致性 如信校度：

一、信度

1、信度（reliability）即可靠性，它指的是采取同样的方法对同一对象重复进行测量时，其所得结果相一致的程度。从另一方面来说，信度就是指测量数据的可靠程度。

2、信度是指测验结果的一致性、稳定性及可靠性，一般多以内部一致性来加以表示该测验信度的高低。信度系数愈高即表示该测验的结果愈一致、稳定与可靠。

3、系统误差对信度没什么影响，因为系统误差总是以相同的方式影响测量值的，因此不会造成不一致性。反之，随机误差可能导致不一致性，从而降低信度。

二、效度

1、效度（Validity）即有效性，它是指测量工具或手段能够准确测出所需测量的事物的程度。效度是指所测量到的结果反映所想要考察内容的程度，测量结果与要考察的内容越吻合，则效度越高；反之，则效度越低。效度分为三种类型：内容效度、准则效度和结构效度。

2、效度是测量的有效性程度，即测量工具确能测出其所要测量特质的程度，或者简单地说是指一个测验的准确性、有用性。效度是科学的测量工具所必须具备的最重要的条件。

3、在社会测量中，对作为测量工具的问卷或量表的效度要求较高。鉴别效度须明确测量的目的与范围，考虑所要测量的内容并分析其性质与特征，检查测量的内容是否与测量的目的相符，进而判断测量结果是否反映了所要测量的特质的程度。

C. Shape

e.g. normality test, normal distribution, random distribution, exponential distribution, bell curve. Shape provide conclusions about overall trends and existence of sub-segments of data.

数据是否服从某种分布

D. Spread

e.g. range, quartiles, interquartile range, variance, standard deviation (e.g. boxplot, σ, test for equal variance). Spread provide conclusions about variation of both Y and X as well as how consistent and homogeneous performance is.

数据的分布情况

E. Central tendency

e.g. mean, median, mode, average (T tests, Anova). Central tendency provide conclusions about the average, to what extent performance is on or off target, and performance differences between different initiatives, artifacts, people, processes, or services.

数据的趋势

F. Capability

e.g. performance tests vs. expectations/targets (Z, USL, LSL, defect, performance gap) . Capability test provide conclusions about the overall performance of the data studied vs. the objectives/targets, and to which extent gaps and improvement options exist.

数据和目标的差距