void clear()
{
_table.clear();
_columns = _rows = 0;
_rules.clear();
_captures.clear();
}
int processData(attrtable& at, datatable& dt, rules& out_rules, int& out_attr_idx)
{
// Rules
int max_idx = -1;
int max_correct = -1;
// For each attribute
for (int i = 0; i < (int)at.size() - 1; ++i)
{
// For checking unique options for each attribute
std::set<std::string> unique;
ruleboard errorboard;
// For each value of that attribute
// Count how often each class appears
for (auto value : dt[i])
{
// C++11 23.4.4.3 map element access [map.access]
// If no key found, the int value is zero initalized
++errorboard[dt[dt.size() - 1][value.first]][value.second];
unique.insert(value.second);
}
// Find the most frequent class
rules cur_rule;
int total_correct= 0;
for (auto value : unique)
{
int max = -1;
std::string classification;
for (auto e : errorboard)
{
if (e.second[value] > max)
{
max = e.second[value];
classification = e.first;
}
}
total_correct += max;
// Make the rule assign that to this attribute value
cur_rule.emplace(std::string(value), std::string(classification));
}
// Choose the rules with largest correct rate (same as smallest error rate)
if (max_correct < total_correct)
{
max_idx = i;
max_correct = total_correct;
out_rules.clear();
out_rules = rules(cur_rule);
}
}
// We found the best attribute that has largest correct rate,
// return the count of correct
out_attr_idx = max_idx;
return max_correct;
}
