func LogisticRegression(c *gopark.Context) {
type dataPoint struct {
x gopark.Vector
y float64
}
points := c.TextFile("points.txt").Map(func(line interface{}) interface{} {
vs := strings.Fields(line.(string))
vector := make(gopark.Vector, len(vs)-1)
for i := 0; i < len(vs)-1; i++ {
vector[i], _ = strconv.ParseFloat(vs[i], 64)
}
y, _ := strconv.ParseFloat(vs[len(vs)-1], 64)
return &dataPoint{vector, y}
}).Cache()
hx := func(w, x gopark.Vector, y float64) float64 {
return 1/(1+math.Exp(-1*w.Dot(x))) - y
}
var w gopark.Vector = []float64{1, -10}[:]
for i := 0; i < 10; i++ {
gradient := points.Map(func(x interface{}) interface{} {
p := x.(*dataPoint)
return p.x.Multiply(-1 * hx(w, p.x, p.y))
}).Reduce(func(x, y interface{}) interface{} {
return x.(gopark.Vector).Plus(y.(gopark.Vector))
}).(gopark.Vector)
w = w.Minus(gradient)
}
fmt.Println("Final Weights:", w)
}
func WordCount(c *gopark.Context) {
txt := c.TextFile("../")
counts := txt.FlatMap(func(line interface{}) []interface{} {
vs := strings.Fields(line.(string))
words := make([]interface{}, len(vs))
for i := range vs {
words[i] = vs[i]
}
return words
}).Map(func(x interface{}) interface{} {
return &gopark.KeyValue{x, 1}
}).ReduceByKey(func(x, y interface{}) interface{} {
return x.(int) + y.(int)
}).Cache()
fmt.Println(
counts.Filter(func(x interface{}) bool {
return x.(*gopark.KeyValue).Value.(int) > 50
}).CollectAsMap())
fmt.Println(
counts.Filter(func(x interface{}) bool {
return x.(*gopark.KeyValue).Value.(int) > 50
}).Map(func(x interface{}) interface{} {
keyValue := x.(*gopark.KeyValue)
keyValue.Key, keyValue.Value = keyValue.Value, keyValue.Key
return keyValue
}).GroupByKey().Collect())
}
func ComputePi(c *gopark.Context) {
N := 100000
iters := c.Data(make([]interface{}, N))
count := iters.Map(func(_ interface{}) interface{} {
x := rand.Float32()
y := rand.Float32()
if x*x+y*y < 1 {
return 1
} else {
return 0
}
}).Reduce(func(x, y interface{}) interface{} {
return x.(int) + y.(int)
}).(int)
fmt.Println("Pi =", (4.0 * float64(count) / float64(N)))
}
