To Understand the ML pipeline in an Interactive way check out the Explorer App.

Get the Status of Data

Visualize The Data

Plot the histograms and know the scaling of your data data.hist(bins=10,figsize = (_,_),grid= False, Etc......)
Visualize the importance of each feature on the target variable sns.FacetGrid(data, hue= "Column_name", col = "column_name", row, etc...) and plot either hist or scatter..
You can use sns.factorplot() for some variables
sns.boxplot(x, y, data)
Plot the distributions if varibale has different classes by defining kind='kde'
Plot the Heat Map of all the columns to get a better view corr = data.corr() and sns.heatmap(corr, .....)
Plot some Vilions on the way

You can either replace with mean or most repeating value of in the column or know the columns that are impacting the missing columns and replace them with their respective similar column values

Taking care of missing categorical classes by assigning a new class to it.
Combining similar columns to a single column.
Binning a column into groups
converting categorical into numerical variables such as count of the letters in a word, on hot encoding, LabelEncoder etc..
Triming down numercial columns for better understading.
Fill Null values by modeling the data on a ml algorithm with null values as test data and the remaining as train data.
Scaling all the numerical fetures into a single scale

Train a small portion of the data and claculate the scores on different algorithms to find a significant increase or decrease in the scores
Use Kflod validation of determine a good model
Get the important features that are helping to preidct the data better
Filter out the important features and use tecniques like Gradient Boosting, AdaBoost Etc.
Train the best models on whole data.
For Better Predictions Ensemble the best models using Maximum Voting or percentages Etc.
Submit the Predictions OR Test the model on the Test Data.