Preface: Completely do not understand the data analysis, statistics also forget the almost small white began to learn data analysis. Read the "In-depth data analysis", the data analysis has a general understanding. Every chapter in the book requires a lot of information to continue learning. This book is a guideline (index).
1. Process of data analysisIdentify goals or issues----decompose the data--to assess the problem, summarize the conclusions--the idea of guiding decision-making data analysis is so, so does the data analysis report.
2. ExperimentsExperiments can help with analysis. To be added during the experiment
control Group, it is easy to find the conclusion of experiment. Use
Random SelectionThe control group is a better approach.
3. OptimizationThe optimization problem mainly consists of three parts:
decision variables, constraints and objective functions。
4. Data visualizationData visualization is a better way to identify problems in the data analysis process and to better present problems or conclusions in data reporting. Scatter plot: Used to show the causality of the two variables, in fact, the scatter plot can only show the correlation of the two variables, the cause and effect of the need to use other things to analyze. (A hollow circle is a better way to represent overlapping relationships.) ) Multivariate scatter plot: The relationship of multiple variables. Histogram: Shows the distribution of the data. 5. Hypothesis test (not understood) 6. Bayesian statistics (not understand) related topic: Basic probability and wave mathematics. 7. Subjective probability (not understood)
Standard deviationEvaluation data. The subjective probability is corrected by Bayesian. 8. Heuristics (not understood)
9. Regression and predictionRegression plus control experiments can predict the future. Regression line: Straight lines that run through the average, can be evaluated using correlation coefficients. The regression line is linear and non-linear.
10. Reasonable errorThe error range allows the user not only to know the predicted value but also to know the error of the norm, making the forecast more credible. It is important to pay attention to the threshold range of the data in the forecast process, and the forecast exceeding the threshold range is inaccurate.
root mean square errorEvaluate the accuracy of forecasts. Pass
SegmentedPrediction and evaluation can control errors.
11. Data CollationExcel and
Regular ExpressionsVery useful. After finishing the data, you need to review the data repeatability and so on.
12. Appendix (Tell me what else to see)1) Statistics 2) Excel3 Yale professor Edward Tufte Graphic principles 4) non-linear and multivariate regression 5) original hypothesis--optional hypothesis reference "in layman statistics" 6) randomness 7) Google Docs can draw and access real-time database 8) Professional skills
Data analysis at a Glance 01: Read the analysis of in-depth data
