For example, I have this array and calculate mean of rows: <pre class="prettyprint"><code>a = np.array([[1,2,3],[2,np.NaN,4]]) mins = np.min(a, axis = 1) </code></pre> The problem is the output is: <code>[1. nan]</code>. How to ignore nan in a and get result <code>[1 2]</code>?

Another more concise and slightly faster alternative is to use the <code>numpy.nanmin()</code> function, which does exactly what you asked for.

You can use masked arrays <pre class="prettyprint"><code>mins = list(np.min(np.ma.masked_array(a, np.isnan(a)), axis=1)) </code></pre>

Calculate min of rows ignoring NaN values

For example, I have this array and calculate mean of rows:

a = np.array([[1,2,3],[2,np.NaN,4]])

mins = np.min(a, axis = 1)

The problem is the output is: [1. nan]. How to ignore nan in a and get result [1 2]?

Another more concise and slightly faster alternative is to use the numpy.nanmin() function, which does exactly what you asked for.

You can use masked arrays

mins = list(np.min(np.ma.masked_array(a, np.isnan(a)), axis=1))