The current situation of measurement of data analysis of Product Manager website (III.)

Cough ~ Plus fill, write more today. hehe ~

This chapter will start to involve the product Manager data collection Mathematical Statistics method ~

Body:

We talked about key quality features "Links: How product managers quantify key requirements indicators"

We understand that key quality features are an important indicator of driving customer satisfaction, and how to improve key quality features, then an indicator is needed for ascension.

After finding the key quality features, we use brainstorming, subdivision tree and other analytical tools to find out the process indicators and process input indicators that affect the key quality characteristics. Measure according to this indicator.

For example, we above "link: Product Manager website Data analysis measurement problem status (b)" The final picture is the customer order product delivery process.

The approximate process is as follows:

Order payment completed 1--according to the product type and delivery area, the order sent to the corresponding distribution personnel 2--according to the order of product packaging, after the completion of the logistics company 3--for product delivery 4-- Customer received product satisfaction rating 5

The output units involved are:

1: Customer

2: Order Processing Staff

3: Product distribution Personnel

4: Logistics Company

We use order Processing, product delivery and product delivery as a measurement target, and then identify the key elements that will affect the mailing of the product and address it as a key optimization. Only when the key link is handled well, can effectively reduce the overall mailing time.

After the measurement indicators are identified, the data collection criteria for the indicators need to be clarified and team formulas formed to ensure the accuracy and consistency of the data measurements. For example, for the product mailing time, whether to start the calculation after the customer submits the order, or from the product to the logistics company began to calculate, this needs to be clear, otherwise it will be due to the collection of data standards are not uniform resulting in data unavailability.

It is important to note that when we determine the measurement indicators, we should pay attention to the difference between continuous data and discrete data in the measurement indicators.

The so-called continuous data refers to data that can be infinitely divided, such as time, length, temperature, their basic unit is linear and continuous.

discrete data refers to individual, individual, or non-overlapping categories of data, such as eligibility/nonconformity, satisfaction rating, gender, etc. In general, continuous data can provide more information about the process and require fewer samples than discrete data. On the statistical tools side, continuous data can provide more analytical tools support. Therefore, when choosing the measurement index, we should try to select continuous data for measurement.

When we have a clear plan for the measurement indicators, we need to determine the measurement object next.

In general, it is not necessary to collect all object data, one is to measure all objects because the workload is too large to measure the cost of data measurement, and second, many times the measurement of all objects is not practical, such work can hardly be completed. The more commonly used method is to select a certain sample size, through the sample data to infer the overall data, this can not only meet the overall analysis, but also improve work efficiency, reduce unnecessary work waste.

The key to sampling is to ensure that the sample is representative, and that the sample size meets the need to correctly estimate the overall characteristics. Common sampling methods include random sampling and stratified sampling.

Random sampling: Extracting samples directly from the whole according to stochastic principle;

Stratified sampling: First, the population in accordance with a certain number of layers, and then in each layer by a simple random sampling sample extraction, each layer of the sample volume is usually the size of the overall sample of the layer is proportional.

In data analysis, the two most commonly used data analysis tools are:

1 Average , the average is expressed in μ.

The concentration of the average reaction sample, such as 4 children, is 4 6 8 10, the average is 7, it can be said that children age at the point of 7 years of concentration.

2 Standard deviation, the standard deviation is expressed by σ

If accurate values are obtained at the 1/n, it is recommended to use 1/(N-1)

Compared with the former, the latter can more accurately reflect the dispersion of the whole, but use the former to calculate the dispersion degree is reduced.

According to the calculation, σ=2.58

Of course, the size of the fluctuations needs to be compared.

___________________________________________________________________

Handwriting is not easy, and read and cherish, reproduced please specify the source: Simonk @http://www.cnblogs.com/simonk/

More Time:

Monday to Friday daily One more, time 22:00-23:40 between

Saturday/Sunday/non-regular holidays

Welcome to subscribe!

Next forecast: two more data analysis tools

___________________________________________________________________