Lesson 01-08 - Optimizers - Building a Parameterized Model

● What is an optimizer?
○ Find minimum values of functions
○ Build parameterized models based on data
○ Refine allocations to stocks in portfolios
● How to use an optimizer?
○ Provide a function to minimize
○ Provide an initial guess
○ Call the optimizer
● Python example
○ Import pandas, matplotlib, and numpy as normal
○ Import scipy.optimize as spo
○ Def f(x):
■ Y = (X-1.5) **2 + 0.5
■ Print “X = {}, Y = {}”.format(X, Y)
■ Return Y
○ Def test_run():
■ Xguess = 2.0
■ Min_result = spo.minimize(f, Xguess, method = ‘SLSQP’, options = {‘disp’:
True})
■ Print ‘Minima found at: ‘
■ Print “X = {}, Y = {}”.format(min_result.x, min_result.fun)
● Convex problems
○ A real value function f(x) defined on an interval is called convex if the line
segment between any two points on the graph of the function lies above the
graph
○ Steps
■ Choose two points and draw a line between them
■ Convex if the line is above the graph when connecting the two points
■ Also has to have one local minima which is also the global minima
■ Cannot have any flat regions
○ Convex problems are easier to find the minima for than non convex problems
○ Can also find the minima of multidimensional graphs
● Building a parameterized model
○ If you are trying to find a line of best fit, you want to minimize the error (meaning
the distance from each of the points to the line of best fit)
○ You have to find the correct values for m and b in the y=mx+b formula in order to
do this
○ A good error metric would be one of the following
■ Σ abs(ei)
■ Σ (ei)2
■ Simply summing up the errors doesn't work, as some of them may be
negative so taking the absolute value or squared error solves that
problem.
● Python code:
○ Import pandas, matplotlib, numpy, and scipy.optimize as normal
○ Def error(line, data): -> error function
■ Subtract estimate from the line that we are looking at at a given point from
the value of the actual data at each point and then square it (Metric: sum
of squared y axis differences)
● Err = np.sum((data[:, 1] - (line[0] * data[:, 0] + line[1])) ** 2)
● Return err (returns error as a single real value)
○ Def fit_line(data, error_func):
■ Generate initial guess for line model
● L = np.float32([0, np.mean(data[:,1])])
○ Slope = 0, intercept = mean y values
■ Plot initial guess (optional)
● X_ends = np.float32([-5, 5])
● plt.plot(x_ends, l[0] * x_ends + l[1], ‘m--‘, linewidth = 2.0, label =
“Initial Guess”)
■ Call optimizer to minimize error function
● Result = spo.minimize(error_func, l, args = (data,), method =
‘SLSQP’, options = {‘disp’: True})
● Return result.x
○ Def test_run():
■ Define original line
● L_orig = np.float32([4,2])
● Print “Original Line: C0 = {}, C1 ={}”.format(l_orig[0], l_orig[1])
● Xorig = np.linespace(0,10,21)
● Yorig = l_orig[0] * Xorig + l_orig[1]
● plt.plot(Xorig, Yorig, ‘b ‘, linewidth = 2.0, label = “Original Line”)
■ Generate Noisy data points
● Noise_sigma = 3.0
● Noise = np.random.normal(0, noise_sigma, Yorig.shape)
● Data = np.asarray([Xorig, Yorig + noise]).T
● plt.plot(data[:, 0], data[:, 1], ‘go’, label = “Data Points”)
■ Try to find a line to fit this data
● L_fit = fit_line(data, error)
● Print “Fitted Line: C0 = {}, C1 ={}”.format(l_fit[0], l_fit[1])
● plt.plot(data[:,0], l_fit[0] * data[:,0] + data[:,1], ‘r--‘, linewidth = 2.0,
label = “Fitted Line”)
■ Add a legend a show the plot
● Python code for polynomials
○ Import pandas, matplotlib, numpy, and scipy.optimize as normal
○ Def error_poly(C, data): -> error function
■ Subtract estimate from the line that we are looking at at a given point from
the value of the actual data at each point and then square it (Metric: sum
of squared y axis differences)
 ● Err = np.sum((data[:, 1] - np.polyval(C, data[:, 0])) ** 2)
● Return err (returns error as a single real value)
○ Def fit_poly(data, error_func, degree = 3):
■ Generate initial guess for polynomial model (all coeffs = 1)
● Cguess = np.poly1d(np.ones(degree+1, dtype = np.float32))
■ Plot initial guess (optional)
● X_ends = np.float32([-5, 5])
● plt.plot(x, np.polyval(Cguess, x), ‘m--‘, linewidth = 2.0, label =
“Initial Guess”)
■ Call optimizer to minimize error function
● Result = spo.minimize(error_func, l, args = (data,), method =
‘SLSQP’, options = {‘disp’: True})
● Return np.poly1d(result.x)
○ Everything else is the same as you would do for a fitted line
● You can use this
○ On functions besides polynomials
○ To model stock prices
○ To Optimize a Portfolio