fdabrandao / mo-book-with-ampl Goto Github PK

I am not able to run LAD-regression (chapter 2), it seems that something broke with some amplpy or numpy update.
The model is very simple:

# indexing sets
set I;
set J;

# parameters
param y{I};
param X{I, J};

# variables
var ep{I} >= 0;
var em{I} >= 0;
var m{J} >= 0;
var b;

# constraints
s.t. residuals {i in I}:
    ep[i] - em[i] == y[i] - sum{j in J}(X[i, j] * m[j]) - b;

# objective
minimize sum_of_abs_errors: sum{i in I}(ep[i] + em[i]);

To call the model

def lad_regression(X, y):
    ampl = AMPL()
    ampl.read("lad_regression.mod")

    n, k = X.shape

    # note use of Python style zero based indexing
    ampl.set["I"] = list(range(n))
    ampl.set["J"] = list(range(k))

    ampl.param["y"] = y
    ampl.param["X"] = X

    ampl.option["solver"] = SOLVER
    ampl.solve()

    return ampl


m = lad_regression(X, y)
m.display("m")
m.display("b")

The issue shows up when X is assigned to ampl.param["X"], X is a npdarray and comes from:

X, y = make_regression(n_samples=n_samples, n_features=n_features, noise=noise)

Use Ipopt, Mosek, Gurobi, Knitro to solve SOCP and exponential cones (Gurobi only SOCP). Progress status for text only:

• Economic order quantity
• The Kelly Criterion
• Markowitz portfolio optimization problem revisited: Mosek fails
• Optimal design of multi-layered building insulation
• Training Support Vector Machines with Conic Programming
• Extra material: Luenberger’s Investment Wheel
• Extra material: Optimal Growth Portfolio

Progress status for models:

• Economic order quantity
• The Kelly Criterion
• Markowitz portfolio optimization problem revisited: Mosek fails
• Optimal design of multi-layered building insulation
• Training Support Vector Machines with Conic Programming
• Extra material: Luenberger’s Investment Wheel
• Extra material: Optimal Growth Portfolio

Notebooks:

• Milk pooling and blending
• Ordinary Least Squares (OLS) Regression
• Markowitz portfolio optimization
• Support Vector Machines for Binary Classification
• Extra material: Refinery production and shadow pricing with CVXPY
• Extra Material: Cutting Stock

Progress for text only:

• Fleet assignment problem
• Robustness analysis of BIM production plan via simulations

Progress for models:

• Fleet assignment problem
• Robustness analysis of BIM production plan via simulations

intro.md needs more adaption, like authors, etc