Expected number of comparisons will be the sum of the product of probability of an item being the searched value and the no. of comparisons for the same. We are given that search key is chosen randomly from one of the keys present in the tree. So, the probability for each item being the searched key = $\frac{1}{n}$, where $n$ is the total no. of keys in the tree.
In a BST, if an item matches at level $h$, we would have done $h$ comparisons.
So, $\text{Expected no. of comparisons} = \sum_{i = 1}^n h_i \times \frac{1}{n} \\ \text{where }h_i \text{ is the level of node }i \\ =\sum_{i=1}^7 h_i \times \frac{1}{7} \\ = \frac{1}{7} + \frac{2}{7} + \frac{2}{7} + \frac{3}{7} + \frac{3}{7} + \frac{3}{7} + \frac{4}{7} \\ = \frac{18}{7} = 2.57$.