def compute_info_gain(data, att_idx):
    info_gain = 0.0
    # = compute_entropy(total) - SUM(branch/attr_total*(compute_entropy(branch)))
    ########## Please Fill Missing Lines Here ##########
    l = self.attrs[att_idx]
    d = {}
    t = 0
    s = 0
    d_keys ={}
    for i in l:
        if i in d:
            d[i]=d[i]+1
            d_keys[i].append(i)
        else:
            d[i]=1
            d_keys[i] = []
            d_keys[i].append(i)
        t+=1
    for i in range(len(l)):
        if l[i] in d:
            d_keys[l[i]].append(i)
        else
            d_keys[l[i]] = []
            d_keys[l[i]].append(i)
    for x in d:
        frac = d[x]/t
        branch = data.get_subset(data, d_keys[x])
        s += frac*(compute_entropy(branch))
    info_gain = compute_entropy(data) - s
    return info_gain