Divide a calculated value/difference by the prior value -> 'slope'

Question

I have a pandas DataFrame df (gradient is the new calculated column):

Time Value1 Value2  gradient    
1        20     40       NaN    
2        30    100       0.5    
3        36    150       0.2    
4        18    100      -0.5

Now, I want to calculate all differences between the row y and the prior row x, divided by the value in the prior row.

Example: in this case, in the new column df['gradient'] should I get in row 3 for 'value1' : 0.20 --> ((36-30) / 30)

I have now:

df['gradient'] =  df['Value1'].diff() / df['Value1']

But I know the / df['Value1'] is not correct. What's the right syntax in this case for dividing by the value in the prior row?

Answer 1

You can shift() the column that you divide by. This moves each value into the next row down. For example if you have your df as:

   Value1
0      20
1      30
2      36
3      18

Then the gradients can be calculated with the division:

>>> df.Value1.diff() / df.Value1.shift()
0    NaN
1    0.5
2    0.2
3   -0.5