what do I do if I have zero values in my data set and want to perform this transformation for a positive skew? It keeps telling me a division of zero was attempted and that the dataset set this a system-missing value
I have a dataset of 16 parameter in 175 water sample. Except one parameter all 15 parameters are skewed. Some get normalized by logarithmic while some don't. To the remaining I performed square root transformation. Now I am confused. Will the result will be valid of Finding a correlation among all the normalized parameters (log vs sq root)?
Thank you for the explanation. It was great. I have a question though... in my transformed data I am having that same issue. After I have transformed with Log10 the skewness shows a normal distribution but either KS or Shapiro-Wilk doesn't. I proceed with Ln and square root transformation and the same is happening. I am dealing with a two-way anova. Some groups end up being normal distributed but some others remain non-normal. Can I choose my independent variables (resulting with normal distribution) and report them under different transformations? I.e., group A was significantly different with a normal distribution (p value of 0.035) after being transformed with log10 while group B was significantly distributed with a square root transformation (p value of 0.012)?? Is there a video where I can see an example of this? Thanks so much for the explanation here!! :)
A very informative video but I have a simple doubt. So if there was another column say "Neutral Skewness" and it had a normal distribution, can we run the the entire dataset when the other two columns i.e. +ve and -ve are normalized. Or should we have to sqrt the third column also to maintain a equality. Thank you
is .72 an acceptable skewness? I found somewhere that a skew value between .5 and 1 is moderate and that the normality is between 0 and 0.5.
what do I do if I have zero values in my data set and want to perform this transformation for a positive skew? It keeps telling me a division of zero was attempted and that the dataset set this a system-missing value
I have a dataset of 16 parameter in 175 water sample. Except one parameter all 15 parameters are skewed. Some get normalized by logarithmic while some don't. To the remaining I performed square root transformation. Now I am confused. Will the result will be valid of Finding a correlation among all the normalized parameters (log vs sq root)?
Thank you for the explanation. It was great. I have a question though... in my transformed data I am having that same issue. After I have transformed with Log10 the skewness shows a normal distribution but either KS or Shapiro-Wilk doesn't. I proceed with Ln and square root transformation and the same is happening. I am dealing with a two-way anova. Some groups end up being normal distributed but some others remain non-normal. Can I choose my independent variables (resulting with normal distribution) and report them under different transformations? I.e., group A was significantly different with a normal distribution (p value of 0.035) after being transformed with log10 while group B was significantly distributed with a square root transformation (p value of 0.012)?? Is there a video where I can see an example of this? Thanks so much for the explanation here!! :)
I also have the same problem with multiple regression analysis. Did you solve this problem?
5:40 here
on the interpretation of coefficient after finished the regression, is there any influence of using transformed square root data?
A very informative video but I have a simple doubt. So if there was another column say "Neutral Skewness" and it had a normal distribution, can we run the the entire dataset when the other two columns i.e. +ve and -ve are normalized. Or should we have to sqrt the third column also to maintain a equality. Thank you
is it necessary to correct kurtosis for normalization?
Great info