Polyfit of stock value

A simple demo of an interactive polyfit extrapolation of stock values

• Features

  • Graphics quibs

  • Graphics-driven assignments

  • Quib-linked widgets

  • Inverse assignments

• Try me

  • Drag the cyan marker to move the fit period.

  • Drag the dashed lines to change the width of the fit period.

  • Use the slider to change the polynomial degree of the fit.

from pyquibbler import iquib, initialize_quibbler, q
initialize_quibbler()

import matplotlib.pyplot as plt
import numpy as np
from matplotlib.widgets import Slider
from datetime import datetime
%matplotlib tk

# Load data of Apple stock:
t_init = datetime.strptime('2000-01-01', '%Y-%m-%d').timestamp()
str2date = lambda x: (datetime.strptime(x.decode("utf-8"), '%Y-%m-%d').timestamp()
                      - t_init) / (365 * 24 * 60 * 60) + 2000
filename = iquib('AAPL.csv')
d = np.genfromtxt(filename, delimiter=',', names=True, converters = {0: str2date})

# Initiate a figure:
plt.figure()
plt.axis([2017, 2022, 0, 200])

# Plot the stock values:
t = d['Date']
y = d['Open']
plt.plot(t, y, 'g', lw=1)

# Define fit parameters:
t_final = iquib(2020.)
dt = iquib(2.)
n = iquib(2)

# define t_initial, starting with dt so that changes in t_initial affects dt
t_initial = -dt + t_final

# Define downstream function quibs of the fitted period:
fit_range = (t >= t_initial) & (t <= t_final)

# Do the fit and plot
p = np.polyfit(t[fit_range], y[fit_range], n)
plt.plot(t[fit_range], np.polyval(p, t[fit_range]), 'r-')

# Plot extrapulation:
extrapulate_range = t > t_final
plt.plot(t[extrapulate_range], np.polyval(p, t[extrapulate_range]), 'r:')

# Plot the fit parameters:
a11 = np.array([1, 1])
plt.plot(t_final * a11, [0, np.polyval(p, t_final)], 'k--', lw=1)
plt.plot(t_initial * a11, [0, np.polyval(p, t_initial)], 'k--', lw=1)

# Add the slider for the polynomial degree:
ax = plt.axes([0.3, 0.8, 0.3, 0.03])
Slider(ax=ax, valmin=0, valmax=3, valinit=n, valstep=1, label='degree');