Kim 2 ML

Starting off as a muggle that naïve to the Math's and Data Science world.

Blog

  • Multivariate Methods for Data Analysis

    Day 68 Type of Multivariate Techniques Day 69 Statistical Significance and PowerPowerMultiple RegressionStep to form Regression Day 70 Tutorial Day 71 Factor Analysis Day 72 Tutorial Day 73 Cluster AnalysisK-means Cluster Day 74 Tutorial Day 75 Multiple Discriminant Analysis Day 76 Tutorial

  • Day 76

    Tutorial 4 Question 1 Example Answer

  • Day 75

    Day 75

    Multiple Discriminant Analysis ps. Classify observation into groups (non metric). SPSS Step 1. Normality of the independent variable distribution test result. Note that KS test is meant for n > 50 and SW test is meant for n < 50. Sig value > 0.05 means accept null hypothesis (variable are normally distributed). Step 2. Collinearity…

  • Day 74

    Tutorial 3 Question 1 Example Answer

  • Day 73

    Day 73

    Cluster Analysis ps. Group similar objects/observations based on the characteristic in a cluster that different to other object in other cluster. Also known as Q analysis or taxonomy. Hierarchical Cluster SPSS Step 1. Bring up Cluster analysis dialog box. Step 2. Configure all cluster setup. Step 3. Interpret agglomeration schedule and dendrogram. Look into Cluster…

  • Day 72

    Tutorial 2 Question 1 Example Answer Exclude Q7 and Q8 Exclude Q6 as it is cross loading, show almost similar magnitude in factor 1 and factor 2. Factor 1 can be named as “Eating habits”.Factor 2 can be named as “Food preparation”.Surrogate variable = Q1 for factor 1, Q10 for factor 2.Summated scale f1 =…

  • Day 71

    Day 71

    Factor Analysis ps. Not looking the effect of 1 variable to another variable. Only looking at how to group variable together that form a common characteristics. Underlying structure also known as factor. SPSS Step 1. Bring up Factor analysis dialog box. Step 2. Configure all factor setup. Step 3. Correlation test result. Relationship between 2…

  • Day 70

    Tutorial 1 Question 1 Example Answer Question 2 Example Answer Exclude Pulse Exclude BSA By using stepwise method, independent variable are introduce/remove step by step while training the model.

  • Day 69

    Statistical Significance and Power ps. This is to explain the consequences of making wrong judgement on statistical test. H0 is True H0 is False Accept H0 Type 2 Error Reject H0 Type 1 Error Power ExampleType I error (false positive) involves wrongly diagnosing a healthy person with a medical condition, risking unnecessary surgery. Type II…

  • Day 68

    Day 68

    Type of Multivariate Techniques

  • Time Series Analysis and Forecasting

    Day 55 Objective of time seriesTime Series Chart (Excel)Time Series Chart (R) Day 56 Time Series componentsStationary TestQualitative Model vs Quantitative ModelTutorial 1 Day 57 Smoothing Techniques Day 58 Smoothing Techniques (Con’t) Day 59 Tutorial 2 Day 60 Measuring Predictive Accuracy Day 61 Box Jenkins Methodology Day 62 Box Jenkins Methodology (Con’t) Day 63 Tutorial…

  • Day 65

    Tutorial 5 Question 1 A volatile model is fitted to a stock return data and produce the following results (a) Write down the equation to forecast volatility based on the output above. (b) What can we conclude based on the results in ARCH LM-test? (c) Are all the coefficients in the model significant? Explain your…

  • Day 64

    Volatile Model (ARCH & GARCH) Technique to address irregular fluctuation that cannot be handle by typical time series model such as ARIMA, Moving Average, Exponential Smoothing & others. R Import data. (download from https://finance.yahoo.com/quote/005930.KS/history) Check order of differencing. Result: Stationary check after differencing. Result: Check stationary and ARIMA hyperparameter. Result: Train an ARIMA model. Check…

  • Day 63

    Tutorial 4 Question 1 The table below shows the result after fitting a time series data by an ARIMA model. (a) Write the ARIMA model equation based on the results above. (b) Represent the ARIMA model in ARIMA (p, d, q). (c) Is the ARIMA model adequate? Justify your answers. (d) Are all the AR…

  • Day 62

    Box Jenkins Methodology (Recap) Split the below data into training (80%) and testing data (20%). Analyse the training data and formulate the model equation for the ARIMA model you chosen: ​ Then, compute the accuracy of the model in the testing data. ​Check the residuals and test whether the model you chosen is satisfactory. Sales Import data. Split Train/Test. Check stationary and…